BODY WASTING AMONG TUBERCULOSIS PATIENTS IN URBAN UGANDA
by
EZEKIEL MUPERE MBChB, M.MED (Paediatrics), M.S
Submitted in partial fulfillment of the requirements
For the degree of Doctor of Philosophy
Dissertation Advisor: Dr. Daniel Tisch PhD, MPH
Department of Epidemiology and Biostatistics
CASE WESTERN RESERVE UNIVERSITY
May, 2010
i
DEDICATION
To
My wife Harriet Mupere Babikako for your cheerful love, prayers,
and persuasive encouragement
My children Patience Esther Mupere, Elizabeth Peace Mupere, Ednah Precious Mupere,
and those I prayerfully hope to be born, for being a strong encouragement. Desire to
embrace this academic path
My elder brother Daniel Bawunha for establishing the foundation of my academic path
and for your words of wisdom
And to,
The memory of my late mother Mary Dorothy Wandawa for the exemplary parental love
and vision, and for introducing me to the objective of learning a, e, i, . . .., and to the objective of learning how to write a love letter; now accomplished. I’m forever indebted.
ii
TABLE OF CONTENTS
LIST OF TABLES ...... vi
ACKNOWLEDGEMENT ...... xx
Abstract ...... xxiii
CHAPTER 1 ...... 1
INTRODUCTION ...... 1
CHAPTER 2 ...... 15
BACKGROUND ...... 15
CHAPTER 3 ...... 32
STUDY METHODS ...... 32
CHAPTER 4 ...... 46
BODY COMPOSITION MEASURED WITH BIOELECTRICAL IMPEDANCE
ANALYSIS AND ANTHROPOMETRY AMONG HIV POSITIVE AND HIV
NEGATIVE ADULTS WITH OR WITHOUT TUBERCULOSIS IN KAMPALA,
UGANDA...... 46
CHAPTER 5 ...... 84
INDICATORS OF DIETARY ADEQUACY AMONG HIV POSITIVE AND HIV
NEGATIVE ADULTS WITH OR WITHOUT TUBERCULOSIS IN URBAN
KAMPALA, UGANDA ...... 84
CHAPTER 6 ...... 126
iii
PREDICTORS OF FAT MASS AND LEAN TISSUE AMONG HIV POSITIVE AND
HIV NEGATIVE ADULTS WITH OR WITHOUT TUBERCULOSIS IN URBAN
KAMPALA, UGANDA ...... 126
CHAPTER 7 ...... 169
BODY WASTING AND DIETARY INTAKE AMONG HIV POSITIVE AND HIV
NEGATIVE ADULTS WITH OR WITHOUT TUBERCULOSIS IN URBAN
UGANDA, KAMPALA ...... 169
CHAPTER 8 ...... 202
CORRELATES OF DIETARY INTAKE AMONG HIV POSITIVE AND HIV
NEGATIVE ADULTS WITH OR WITHOUT TUBERCULOSIS IN URBAN
KAMPALA, UGANDA ...... 202
CHAPTER 9 ...... 233
IMPACT OF BODY WASTING ON SURVIVAL AMONG ADULT PATIENTS
WITH PULMONARY TUBERCULOSIS IN URBAN KAMPALA, UGANDA ..... 233
CHAPTER 10 ...... 292
LONGITUDINAL CHANGES IN BODY COMPOSITION AMONG HIV POSITIVE
AND HIV NEGATIVE ADULT PATIENTS WITH PULMONARY
TUBERCULOSIS IN URBAN KAMPALA, UGANDA ...... 292
CHAPTER 11 ...... 347
SUMMARY OF RESULTS AND GENERAL CONCLUSIONS ...... 347
CHAPTER 12 ...... 365
iv
EXTRA TABLES AND FIGURES FOR CHAPTER FOUR ...... 365
CHAPTER 13 ...... 377
EXTRA TABLES AND FIGURES FOR CHAPTER NINE ...... 377
CHAPTER 14 ...... 414
EXTRA TABLES AND FIGURES FOR CHAPTER TEN ...... 414
BIBLIOGRAPHY ...... 441
v
LIST OF TABLES
Table 4:1 Characteristics of study population with tuberculosis ...... 65
Table 4:2 Characteristics of study population without tuberculosis ...... 67
Table 4:3 Comparison fat-free mass and fat mass as measured by BIA and by equations
that involved anthropometry measurements for all participants (n=131) ...... 69
Table 4:4 Comparison fat-free mass and fat mass measured by BIA and by equation with
waist circumference ...... 70
Table 4:5 Comparison fat-free mass and fat mass measured by BIA and by equation with
sum of 4 skinfolds ...... 72
Table 4:6 Comparison fat-free mass and fat mass measured by BIA and by equation with
BMI ...... 74
Table 4:7 Comparison fat-free mass and fat mass measured by BIA and by equation with
MUAC...... 76
Table 5:1 Characteristics of adult participants with tuberculosis in Kampala, Uganda . 104
Table 5:2 Characteristics of adult participants without tuberculosis in Kampala, Uganda
...... 105
Table 5:3 Food groups and food items from 24-hour dietary intake recall among HIV positive and HIV negative adults in Kampala, Uganda (n=131) ...... 106
Table 5:4 The 24-hour dietary intake recall, recommended daily allowance, frequency of inadequate intake of nutrients among HIV positive and HIV negative adults in Kampala,
Uganda (n=131) ...... 108
vi
Table 5:5 Percent of inadequate 24-hour dietary recall intake among HIV positive and
HIV negative adults with tuberculosis in Kampala, Uganda ...... 110
Table 5:6 Percent of inadequate 24-hour dietary recall intake among HIV positive and
HIV negative adults without tuberculosis in Kampala, Uganda ...... 112
Table 5:7 Spearman’s correlations between Nutrient Adequacy Ratio (NAR) of nutrients and Food Variety Score or Dietary Diversity Score among HIV positive and HIV negative adults with or without tuberculosis in Kampala, Uganda (n=131)...... 114
Table 5:8 Mean MAR scores for different levels of Food Variety Score (FVS) and
Dietary Diversity Score (DDS) among HIV positive and HIV negative adults in Kampala,
Uganda (n=131) ...... 116
Table 5:9 Estimated Mean Adequacy Ratio scores for different levels of Food Variety
Score (FVS) and Dietary Diversity Score (DDS) from Linear Regression Model among
HIV positive and HIV negative adults in Kampala, Uganda (n=131) ...... 118
Table 6:1 Characteristics of the study population (n=131) ...... 145
Table 6:2 Nutrient intake characteristics of the study population (n=131) ...... 148
Table 6:3 Spearman’s correlations between energy or protein intake and body mass index, fat or fat-free mass (n=131) ...... 149
Table 6:4 Predictors of fat-free mass in univariate models among HIV positive and HIV negative adults with or without tuberculosis in Kampala, Uganda (n=131)...... 150
Table 6:5 Predictors of fat-free mass in univariate models among HIV positive and HIV negative adults with or without tuberculosis in Kampala, Uganda stratified by sex
(n=131) ...... 152
vii
Table 6:6 Predictors of fat-free mass in multivariable models among HIV positive and
HIV negative adults with or without tuberculosis in urban Kampala, Uganda (n=131) 154
Table 6:7 Predictors of fat-free mass in multivariable models among HIV positive and
HIV negative adults with or without tuberculosis in Kampala, Uganda stratified according sex (n=131) ...... 155
Table 6:8 Predictors of fat mass in univariate models among HIV positive and HIV negative adults with or without tuberculosis in Kampala, Uganda (n=131)...... 156
Table 6:9 Predictors of fat mass in univariate models among HIV positive and HIV negative adults with or without tuberculosis in Kampala, Uganda stratified according to sex (n=131) ...... 157
Table 6:10 Predictors of fat mass in multivariable models among HIV positive and HIV negative adults with or without tuberculosis in Kampala, Uganda (n=131)...... 159
Table 6:11 Predictors of fat mass in multivariable models among HIV positive and HIV negative adults with or without tuberculosis in Kampala, Uganda (n=131)...... 161
Table 6:12 Predictors of body mass index in univariate models among HIV positive and
HIV negative adults with or without tuberculosis in Uganda (n=131) ...... 163
Table 6:13 Predictors of body mass index in univariate models among HIV positive and
HIV negative adults with or without tuberculosis in Kampala, Uganda stratified according sex (n=131) ...... 165
Table 6:14 Predictors of body mass index in multivariable models among HIV positive and HIV negative adults with or without tuberculosis in urban Uganda (n=131) ...... 167
Table 6:15 Predictors of body mass index in multivariable models among HIV positive and HIV negative adults with or without tuberculosis in Kampala, Uganda (n=131) .... 168
viii
Table 7:1 Select characteristics among HIV positive and HIV negative adults with tuberculosis ...... 186
Table 7:2 Select characteristics among HIV positive and HIV negative adults without tuberculosis ...... 188
Table 7:3 Dietary intake of 24-hour recall among HIV positive and HIV negative adults with tuberculosis ...... 190
Table 7:4 Dietary intake of 24-hour recall among HIV positive and HIV negative adults without tuberculosis ...... 192
Table 7:5 Select characteristics among patients with/without wasting (n=63)...... 194
Table 7:6 Select characteristics among patients with/without severity of clinical tuberculosis ...... 196
Table 7:7 Dietary intake of 24-hour recall among tuberculosis patients with/without body wasting (n=63) ...... 198
Table 7:8 Dietary intake of 24-hour recall among tuberculosis patients with/without severe clinical TBscore (>5) ...... 200
Table 8:1 Characteristics of the study population (n=131) ...... 215
Table 8:2 Correlates of dietary energy intake among women and men ...... 218
Table 8:3 Correlates of dietary protein intake among women and men ...... 221
Table 8:4 Proportions of adult individuals with inadequate dietary intakes of energy, protein, and micronutrients in relation to socio-demographic, HIV, and tuberculosis variables in Kampala, Uganda ...... 223
ix
Table 8:5 Proportions of adult individuals with inadequate dietary intakes of energy,
protein, and micronutrients in relation to socio-demographic, HIV, and tuberculosis
variables in Kampala, Uganda ...... 226
Table 8:6 Correlates of inadequate dietary intakes of key nutrients from multivariate
models among adult individuals in Kampala, Uganda ...... 229
Table 9:1 Definitions of low and normal fat and fat-free mass index values for corresponding body mass index in adults ...... 243
Table 9:2 Mean body mass, fat and fat-free mass indexes; spearman correlations between body mass index and fat or fat-free mass indexes among adult women and men in urban
Uganda ...... 264
Table 9:3 Concordance between low body mass index and low fat-free or fat mass
indexes corresponding to body mass index in assessing wasting among adults in urban
Uganda ...... 265
Table 9:4 Baseline characteristics of pulmonary tuberculosis patients with normal versus
malnutrition ...... 267
Table 9:5 Stratified Analysis of mortality among tuberculosis with normal (≥18.5) and
low (<18.5) body mass index (BMI kg/m2) according to key variables [deaths/number at risk (%)] ...... 270
Table 9:6 Stratified Analysis of mortality among tuberculosis with normal and low fat- free mass index (FFMI kg/m2) according to key variables [deaths/number at risk (%)] 272
Table 9:7 Univariate Analysis of factors associated with survival ...... 274
x
Table 9:8 Relative hazards [HR, 95% confidence intervals (CIs)] for death among
tuberculosis patients with normal compared with patients having low body mass index
(BMI) ...... 277
Table 9:9 Relative hazards [HR, 95% confidence intervals (CIs)] for death among
tuberculosis patients with normal compared with patients having low body mass index
(BMI) stratified according to sex status ...... 279
Table 9:10 Relative hazards [HR, 95% confidence intervals (CIs)] for death among
tuberculosis patients with normal compared with patients having low body mass index
(BMI) ...... 281
Table 9:11 Relative hazards [HR, 95% confidence intervals (CIs)] for death among
tuberculosis patients with normal compared with patients having low body mass index
(BMI) stratified according to HIV status ...... 283
Table 9:12 Relative hazards [HR, 95% confidence intervals (CIs)] for death among
tuberculosis patients with normal compared with low fat-free mass index (FFMI) ...... 285
Table 9:13 Relative hazards [HR, 95% confidence intervals (CIs)] for death among
tuberculosis patients with normal compared with low fat-free mass index (FFMI)
stratified according to sex status ...... 287
Table 10:1 Definitions of low and normal fat and fat-free mass index values for corresponding body mass index in adults ...... 302
Table 10:2 Baseline characteristics of the study population with/without baseline wasting
...... 328
xi
Table 10:3 Unadjusted rate of change for fat-free mass (FFMI), fat mass (FMI), and body mass index (BMI) among pulmonary patients with reduced FFMI, FMI, and BMI in
Kampala, Uganda ...... 330
Table 10:4 Adjusted rate of change for fat-free mass (FFMI), fat mass (FMI), and body
mass index (BMI) among tuberculosis patients presenting with reduced FFMI, FMI, and
BMI in Kampala, Uganda ...... 333
Table 10:5 Adjusted rate of change for fat-free mass (FFMI), fat mass (FMI), and body
mass index (BMI) among tuberculosis patients presenting with reduced FFMI, FMI, and
BMI stratified according to gender in Kampala, Uganda ...... 336
Table 10:6 Adjusted rate of change for fat-free mass (FFMI), fat mass (FMI), and body
mass index (BMI) among tuberculosis patients presenting with reduced FFMI, FMI, and
BMI in Kampala, Uganda ...... 339
Table 10:7 Adjusted rate of change for fat-free mass (FFMI), fat mass (FMI), and body
mass index (BMI) among tuberculosis patients presenting with reduced FFMI, FMI, and
BMI stratified according to HIV in Kampala, Uganda ...... 342
Table 12:12:1 Mean intra-supervisor, mean intra-observer, and mean inter-observer
technical error of measurement prior to data collection ...... 366
Table 12:12:2 Coefficient of reliability prior to data collection ...... 367
Table 12:12:3 Guide for comparison of inter-observer error (Frisancho A.R 1990) ...... 368
Table 13:1 Comparison of key baseline variables across phase II prednisolone study,
household contact study, and Kawempe community health study ...... 378
Table 13:2 Assessing the proportional hazards assumption with a statistical test:
Correlations between ranked failure time and Schoenfeld residuals ...... 380
xii
Table 13:3 Baseline characteristics of pulmonary tuberculosis patients between HIV
negative and HIV positive ...... 381
Table 13:4 Multivariable Relative hazards [HR, 95% confidence intervals (CIs)] for death
among tuberculosis patients with normal compared with patients having low body mass
index (BMI) ...... 384
Table 13:5 Multivariable Relative hazards [HR, 95% confidence intervals (CIs)] for death
among tuberculosis patients with normal compared with patients having low body mass
index (BMI) ...... 386
Table 13:6 Relative hazards [HR, 95% confidence intervals (CIs)] for death among
tuberculosis patients with normal compared with patients having low body mass index
(BMI) ...... 388
Table 13:7 Relative hazards [HR, 95% confidence intervals (CIs)] for death among
tuberculosis patients with normal compared with patients having low body mass index
(BMI) stratified according to Age group ...... 390
Table 13:8 Relative hazards [HR, 95% confidence intervals (CIs)] for death among HIV
positive tuberculosis patients compared with HIV negative patients ...... 392
Table 13:9 Relative hazards [HR, 95% confidence intervals (CIs)] for death among HIV
positive tuberculosis patients compared with HIV negative patients stratified according to
BMI categories ...... 394
Table 13:10 Multivariable Relative hazards [HR, 95% confidence intervals (CIs)] for
death among tuberculosis patients with normal compared with patients having low fat- free mass index (FFMI) ...... 396
xiii
Table 13:11 Relative hazards [HR, 95% confidence intervals (CIs)] for death among
tuberculosis patients with normal compared with low fat-free mass index (FFMI) ...... 398
Table 13:12 Relative hazards [HR, 95% confidence intervals (CIs)] for death among
tuberculosis patients with normal compared with low fat-free mass index (FFMI)
stratified according to HIV status ...... 400
Table 13:13 Relative hazards [HR, 95% confidence intervals (CIs)] for death among
tuberculosis patients with normal compared with low fat-free mass index (FFMI) ...... 402
Table 13:14 Relative hazards [HR, 95% confidence intervals (CIs)] for death among
tuberculosis patients with normal compared with low fat-free mass index (FFMI)
stratified according to age group...... 404
Table 13:15 Relative hazards [HR, 95% confidence intervals (CIs)] for death among HIV
positive tuberculosis patients compared with HIV negative patients ...... 406
Table 13:16 Relative hazards [HR, 95% confidence intervals (CIs)] for death among HIV
positive tuberculosis patients compared with HIV negative patients stratified according to
fat-free mass index (FFMI) categories ...... 408
Table 14:1 Spearman raw correlation matrix for fat-free mass, fat mass, and body mass index over different visit time points ...... 414
Table 14:2 Regression estimates for effects of baseline wasting, lag FFMI measure, and
baseline characteristics on probability of missing ...... 416
Table 14:3 Regression estimates for effects of baseline wasting, lag FMI measure, and
baseline characteristics on probability of missing ...... 418
Table 14:4 Assessing contributions of polynomials, random intercepts, and random
slopes in mixed models ...... 422
xiv
Table 14:5 Assessing Piecewise and polynomial models; random intercepts, and random slopes in mixed models ...... 424
xv
LIST OF FIGURES
Figure 1:1 Working tuberculosis – HIV – Malnutrition Model ...... 6
Figure 3:1 Retrospective cohort study flow ...... 41
Figure 3:2 Cross-sectional study flow diagram ...... 42
Figure 5:1 Sensitivity (sens) and specificity (spec) for different cut-off points of Food
Variety Score (FVS) among women: Mean Adequacy Ratio (MAR) changing from 60 to
75...... 122
Figure 5:2 Sensitivity (sens) and specificity (spec) for different cut-off points of Food
Variety Score (FVS) among men: Mean Adequacy Ratio (MAR) changing from 60 to 75
...... 123
Figure 5:3 Sensitivity (sens) and specificity (spec) % for different cut-off points of Diet
Diversity Score (DDS) among women: Mean Adequacy Ratio (MAR) changing from 60 to 75 ...... 124
Figure 5:4 Sensitivity (sens) and specificity (spec) % for different cut-off points of Diet
Diversity Score (DDS) among men: Mean Adequacy Ratio (MAR) changing from 60 to
75...... 125
Figure 9:1 Survival distribution among adult men presenting with wasted body mass
(BMI <18.5 kg/m2) compared to men with normal body mass in urban Kampala, Uganda
...... 289
Figure 9:2 Survival distribution among adult women with baseline fat-free mass wasting
(FFMI <16.7 kg/m2 for men, <14.6 kg/m2 for women) compared to women with normal fat-free mass in urban Uganda ...... 290
xvi
Figure 9:3 Survival distribution Men compared with Women among HIV positive tuberculosis patients in urban Uganda ...... 291
Figure 10:1 Overall Study flow diagram ...... 345
Figure 10:2 Study flow diagram for BIA data ...... 346
Figure 11:1 Overall conclusions: Tuberculosis – HIV – malnutrition model ...... 359
Figure 12:1 Fat-free mass measured by BIA compared with fat-free mass measured by equation involving waist circumference ...... 369
Figure 12:2 Fat mass measured by BIA compared with fat mass measured by equation involving waist circumference ...... 370
Figure 12:3 Fat-free mass measured by BIA compared with fat-free mass measured by
Durnin & Womersely equations involving sum of 4 skinfold thickness ...... 371
Figure 12:4 Fat mass measured by BIA compared with fat mass measured by Durnin &
Womersely equations that involves sum of 4 skinfold thickness ...... 372
Figure 12:5 Fat-free mass measured by BIA compared with fat-free mass measured by equation involving BMI ...... 373
Figure 12:6 Fat mass measured by BIA compared with fat mass measured by equation involving BMI ...... 374
Figure 12:7 Fat-free mass measured by BIA compared with fat-free mass measured by equation involving MUAC ...... 375
Figure 12:8 Fat mass measured by BIA compared with fat mass measured by equation involving MUAC ...... 376
Figure 13:1 Survival distribution among women with low baseline body mass index
(BMI) (<18.5 kg/m2) compared to women with normal BMI in Uganda ...... 410
xvii
Figure 13:2 Survival distribution among men with low (<16.7 kg/m2 for men, <14.6 kg/m2 for women) baseline fat-free mass index (FFMI) compared to men with normal
FFMI in Uganda ...... 411
Figure 13:3 Survival distribution among Men compared to Women tuberculosis patients in urban Uganda ...... 412
Figure 13:4 Survival distribution among Men compared with Women among HIV negative tuberculosis patients in Uganda ...... 413
Figure 14:1 Individual profiles for fat-free mass over time ...... 426
Figure 14:2 Individual profiles for fat mass index over time ...... 427
Figure 14:3 Individual profiles for BMI over time ...... 428
Figure 14:4 Box plots for fat-free mass index over time ...... 429
Figure 14:5 Box plots for fat mass index over time...... 430
Figure 14:6 Box plots for BMI over time ...... 431
Figure 14:7 Mean profile for FFMI among patients with compared to patients without baseline wasting ...... 432
Figure 14:8 Mean profile for FFMI among men compared to mean profile among women
...... 433
Figure 14:9 Mean profile for FFMI among HIV negative compared to mean profile among HIV positive patients ...... 434
Figure 14:10 Mean profile for FMI among patients with compared to patients without wasting ...... 435
Figure 14:11 Mean profile for FMI among men compared to mean profile among women
...... 436
xviii
Figure 14:12 Mean profile for FMI among HIV negative compared to mean profile among HIV positive patients ...... 437
Figure 14:13 Mean profile for BMI among patients with compared to patients without wasting ...... 438
Figure 14:14 Mean profiles for BMI among men compared to mean profile among women ...... 439
Figure 14:15 Mean profile for BMI among HIV negative compared to mean profile among HIV positive patients ...... 440
xix
ACKNOWLEDGEMENT
Several persons contributed in different ways to the successful completion of this dissertation – to them I am profoundly indebted. I wish to express my deep gratitude to:
The Fogarty AIDS International Training and Research Program (AITRP) grant number
TW000011 at Case Western Reserve University (CWRU) for funding my training and accomplishment of this dissertation under directorship of Prof. Christopher Whalen. I am grateful to the administrative staff of the AITRP grant: Alice Cantini and Grace Sivlar at
CWRU, Robert Kakaire and Xiaoping at University of Georgia for the all the administrative support.
The dissertation committee members: Dr. Daniel J. Tisch, Dr. Mark Schluchter, Dr.
Isabel Parraga, Dr. Catherine Stein, and Dr. Christopher Whalen for the guidance and constructive criticisms that enabled me design, analyze, and write-up this dissertation.
Specifically, Dr. Tisch for not only accepting to chair the committee at such a time when
Dr. Whalen had to leave for Georgia but also for the moral support and guidance in drafting the manuscripts; Dr. Schluchter for the design, statistical skills and knowledge that shaped the dissertation; Dr. Parraga for the design and technical expertise in nutrition that has made me envision nutrition epidemiology as a research career; Dr. Stein for accepting at to be a member on the committee at short notice and for aggressively reviewing all the analysis and write style of the manuscripts in this dissertation; and Dr.
Whalen for not only being my chair and research advisor before his transfer to Georgia
xx
but for also introducing me to critical thinking and tenacity in research. Dr. Whalen, I
will remain forever indebted.
Dr. Harriet Mayanja-Kiiza, Professor at the College of Health Sciences Makerere
University Kampala, Uganda, for the mentorship and for the constructive suggestions and criticisms during the design and write-up of this dissertation.
The study staff: Cissy Nambejja, Hassard Sempeera, Ruth Balaba, and Isaac Nsereko for the support in primary data collection which would have been impossible without their input.
All the staff of the Uganda-Case Western Reserve University Research Collaboration and
all the staff of the Tuberculosis Research Unit (TBRU) at Case Western Reserve
University for collecting and maintaining clean data that enabled me accomplish the
survival and longitudinal aims in this dissertation. Special thanks to LaShaunda Malone
and Allan Chiunda who put together the datasets that were used for analysis; to Denise
Jonhson who performed the data management for the primary cross-sectional data
collection; to Dr. Henry W. Boom, the PI of TBRU, for the constructive criticisms and
for granting me permission to use the data; and to the grant N01-AI95383 and
HHSN266200700022C/ N01-AI70022 from the NIAID that provided funding to TBRU.
xxi
The fellow Fogarty trainees for the period January 2006 to May 2010, for the moral
support, constructive criticisms and suggestions in the development and write-up of this
dissertation.
My wife Harriet Mupere Babikako, for standing with me during the good and difficult times in my academic career.
My elder brother, Daniel Bawunha for being a good source of encouragement and
wisdom in my academic pursuit.
xxii
Body Wasting among Tuberculosis Patients in Urban Uganda
Abstract
By
EZEKIEL MUPERE
Background Although body wasting is a cardinal feature of tuberculosis, its etiology and
management is poorly understood; and its assessment is overlooked in research and in
clinical practice.
Objective We established whether body wasting modifies survival and body composition changes during and after tuberculosis treatment; whether HIV modifies dietary intake
among tuberculosis patients; whether dietary intake differs by wasting and severity of
disease status; and whether dietary intake influences body composition.
Methods Retrospective cohort and cross-sectional designs were employed. Height-
normalized body mass (BMI), fat-free mass (FFMI), and fat mass (FMI) indices and 24-
hour dietary intake recall were measured.
Results Body wasting was associated with reduced survival and the effect differed by
gender. FFMI was found to be a predictor of survival among women whereas BMI was
xxiii
among men. Wasting was associated with substantial linear increase in FFMI, FMI, and
BMI during the first three months but the rate of increase differed by gender and not HIV status. Changes in body composition among men were affected by initial FFMI and BMI, whereas among women by FMI. There were minimal changes in body composition after month 3 and during the one year period after month 12 regardless of the initial body composition, gender, and HIV status. Dietary intake in the study population was monotonous, rich in carbohydrates and deficient in nutrients. Dietary intake at the time of diagnosis was influenced by severity of tuberculosis disease, but not HIV status and in the absence of tuberculosis was influenced by gender. Prediction of body composition by energy and protein intake differed by gender. Energy intake was an important predictor of body composition among women whereas appetite was among men.
Conclusion Results provide theoretical framework to provide targeted nutritional intervention to patients presenting with wasting and patients of female gender. National programs should integrate nutritional health education in the management of tuberculosis.
Nutritional assessment should involve establishment of body composition to identify patients that may be at risk of poor survival. Further evaluation is needed to understand changes in dietary intake overtime and its impact on body composition.
Key words: Tuberculosis; body wasting; survival; body composition; dietary intake
Word count: 343
xxiv
CHAPTER 1
INTRODUCTION
1
INTRODUCTION
Tuberculosis is the most relevant infectious disease worldwide, and its etiological agent,
Mycobacterium tuberculosis, infects one-third of the world population (World Health
Organization. Global tuberculosis control. Epidemiology, strategy, financing. WHO
2009). Globally, 9.2 million new cases and 1.7 million tuberculosis-associated deaths occurred in 2006 (Vitoria et al. 2009). In sub-Saharan Africa, where the prevalence and
incidence of tuberculosis are among the highest in the world, 79% adults with
tuberculosis are also infected with human immunodeficiency virus (HIV) (Lawn and
Churchyard 2009; World Health Organization 2009), resulting in elevated mortality.
Tuberculosis is frequently found at autopsy of acquired immunodeficiency syndrome
(AIDS) patients (Lucas et al. 1994; Lucas et al. 1993), particularly cachectic patients
suggesting that tuberculosis exacerbates the wasting process of HIV-infected people in
Africa.
Tuberculosis and HIV are both independently associated with body wasting (Scalcini et al. 1991; Suttmann et al. 1995; Harries et al. 1988; Kennedy et al. 1996). In tuberculosis-
HIV co-infection, there may be additional metabolic, physical, and nutritional burden, resulting in potential further increase in energy expenditure, malabsorption, and
micronutrient deficiency. Thus, the suggested tuberculosis or co-infection malnutrition
model (Figure 1:1). In this model because of differences in metabolism, gender differences in body composition at presentation among TB patients have been reported
2
(Kennedy et al. 1996; Mupere et al. 2010). There may also be increased production of pro-inflammatory cytokines resulting in breakdown of body lipids and proteins
(Niyongabo et al. 1999; van Lettow, Fawzi, and Semba 2003) in a patient with already compromised nutrient intake because of reduced appetite. Moreover, wasting itself is a cause of immune-deficiency affecting cellular immunity the key host defense against tuberculosis (Chandra 1991; Fernandes G, Jolly C.A, and Lawrence R.A 2006; Gershwin
M.E, Beach R.S, and Hurley L.S 1985). Thus, body wasting may be a risk factor for tuberculosis and tuberculosis might not only worsen the course of HIV-associated immune-depression but also worsen the HIV-associated wasting because recent reports suggest tuberculosis to be the dominant factor in driving the wasting process in co- infection (Mupere et al. 2010; Paton and Ng 2006). These interrelated effects possibly explain the fact that both tuberculosis and wasting are associated with reduced survival during HIV infection (Nunn et al. 1992; Suttmann et al. 1995).
Early detection and timely management of body wasting is important in prevention of morbidity and mortality in tuberculosis. The assessment of body wasting and management is also important during and after tuberculosis treatment to understand the changes in body composition. However up to date, several prior studies particularly in sub-Saharan Africa that assessed body wasting in tuberculosis used body mass index
(BMI) (Zachariah et al. 2002; Kennedy et al. 1996; Mugusi et al. 2009). Yet BMI is insensitive to body fatness, particularly at low BMI, as well as with above normal muscle development (Kyle, Genton, and Pichard 2002; Kyle, Piccoli, and Pichard 2003). This implies that previous studies (Zachariah et al. 2002; Mugusi et al. 2009) might have
3
overestimated mortality due to wasting, and might have failed to reveal precise changes in body composition during and after tuberculosis (Kennedy et al. 1996; Ramakrishnan et al. 1961). Moreover, compared to fat mass or weight in itself, fat-free mass body compartment correlates closely with quality of life, physical functioning, and survival
(Wagner, Ferrando, and Rabkin 2000; Mostert et al. 2000). Bioelectrical impedance analysis (BIA) provides a precise and practical method for clinical assessment of fat and fat-free mass (Kyle, Genton, and Pichard 2002; Kyle et al. 2004); however, its application is still limited to research settings that simple and inexpensive methods are needed particularly in sub-Saharan Africa.
Although body wasting and malnutrition play an important role in the clinical course of patients with tuberculosis and HIV and those with dual infection, nutritional status, nutritional intake, and quality of intake are often overlooked in regular clinical practice and in tuberculosis programs. This has led to paucity of information to characterize nutritional intake and quality of intake, and how the intake influence body composition in patients with or without tuberculosis. Characterization of tuberculosis-associated body wasting and nutrient intake not only contributes to the understanding of the pathophysiology but also the management of body wasting.
This dissertation was designed to fill gaps in understanding survival and changes in body composition using precise measures of nutritional status; in understanding nutrient intake
4
and its association with body wasting and body composition measurements; and how body composition can be assessed using simple existing equations in sub-Saharan.
Goal
The goal of this dissertation was to generate information that could be used to improve survival in tuberculosis patients. This dissertation focused first on understanding simple and inexpensive approaches of assessing body wasting. Second, focused on generating information that could aid in understanding of body wasting and its management, that is whether nutritional factors influence body composition in the face of tuberculosis; and whether body wasting as measured by precise measures of body composition modifies the course of tuberculosis.
5
Figure 1:1 Working tuberculosis – HIV – Malnutrition Model
Tuberculosis HIV Reduced appetite Immune activation Inadequate dietary intake Immune deficiency Altered metabolism: HIV replication Increased cytokine production Anabolic block
HIV-TB Co-infection Increased energy expenditure Malabsorption Increased inadequate intake
Lipid and protein metabolism Wasting of body composition: Men (Mupere et al. 2010) Women Begin with high fat-free mass Begin with low fat-free mass Begin low with fat mass Begin with high fat mass Marked fat-free mass wasting Preserve fat-free mass wasting Fat mass wasted in proportion with FFM Marked fat mass wasting Low body mass index High body mass index Gender differences in micronutrient deficiency not known?
Wasting modifies: Men Women Survival and effect of HIV Survival and effect of HIV? Changes: Fat-free mass, fat mass, BMI? Changes: Fat-free mass, fat mass, BMI? Differences in dietary Intake? Energy and protein intake in prediction of body composition? Independent effects of HIV and tuberculosis on dietary intake?
BMI = body mass index, FFM = fat-free mass
6
SPECIFIC AIMS
Chapter 4: Body composition measured with bioelectrical impedance analysis and
anthropometry among HIV positive and HIV negative adults with or without
tuberculosis in urban Kampala, Uganda.
Specific Aim:
To establish whether simple anthropometric measurements can be used to provide
comparable estimates of body composition to those of bioelectrical impedance analysis.
Hypothesis:
Waist circumference, mid-upper-arm circumference, and body mass index provide similar estimates of body composition to those of bioelectrical impedance analysis.
Chapter 5: Indicators of dietary adequacy among HIV positive and HIV negative adults with or without tuberculosis in urban Kampala, Uganda.
Specific Aim:
To evaluate whether simple counting of food items or food groups are indicators of dietary adequacy in a population of HIV positive and HIV negative adults with or without tuberculosis in urban Kampala, Uganda.
7
Hypothesis:
Simple counting of food items or food groups are predictors of dietary adequacy among
HIV positive and HIV negative adults with or without tuberculosis in urban Kampala,
Uganda.
Chapter 6: Predictors of body composition among HIV positive and HIV negative adults with or without tuberculosis in urban Kampala, Uganda.
Specific Aim:
To establish whether energy and protein intake are predictors of body composition among
HIV positive and HIV negative adults with or without tuberculosis in urban Kampala,
Uganda.
Chapter 7: Body wasting and dietary intake among HIV positive and HIV negative adults with or without tuberculosis in urban Kampala, Uganda.
Specific Aim:
To establish the relationship between body wasting or severity of tuberculosis disease and dietary intake among HIV positive and HIV negative adults with or without tuberculosis in urban Kampala, Uganda.
8
Hypothesis:
HIV modifies dietary intake among tuberculosis patients.
Dietary intake differs by body wasting.
Chapter 8: Correlates of dietary intake and adequacy among HIV positive and HIV
negative adults with or without tuberculosis in urban Kampala, Uganda.
Specific Aim:
To establish dietary correlates of energy and protein intake, and correlates of nutrient inadequacy among HIV positive and HIV negative adults with or without tuberculosis in urban Kampala, Uganda.
Hypothesis:
Tuberculosis and HIV do not affect energy and protein intake
Chapter 9: Impact of body wasting on survival among adult patients with pulmonary tuberculosis in urban Kampala, Uganda.
Specific Aim:
To establish the effect of body wasting as measured by height-normalized fat mass and lean tissue and body mass indices on survival among adult patients with pulmonary tuberculosis in urban Kampala, Uganda.
9
Hypothesis:
Body wasting modifies survival among tuberculosis patients in urban Kampala, Uganda.
Chapter 10: Longitudinal changes in body composition during and after tuberculosis treatment in urban Kampala, Uganda.
Specific Aim: To establish the rate of change for fat mass, fat-free mass, and BMI among tuberculosis patients in urban Kampala, Uganda.
Hypothesis:
Body wasting modifies longitudinal changes in body composition during and after tuberculosis treatment.
10
CHAPTER 2
BACKGROUND
15
BACKGROUND
Nutrition and Tuberculosis
The role of nutrition in tuberculosis leads to consideration of three important questions
(Rao K.N and Gopalan C 1966). Does malnutrition predispose to tuberculosis? Does malnutrition influence or modify the course of established infection? Do nutritional factors influence the response to chemotherapy? In the present dissertation, the focus was to understand the influence of body wasting and malnutrition as measured by precise measures of fat and fat-free mass on survival and changes in body composition during and after tuberculosis treatment. Further, to understand the influence of nutritional factors, specifically nutrient intake and its quality on body composition in the presence or absence of tuberculosis.
Effects of nutritional status on tuberculosis
The association between tuberculosis and malnutrition has long been recognized, as malnutrition predisposes people to the development of clinical disease; and tuberculosis often exacerbates malnutrition (Macallan 1999). The host protective immune mechanism of infection with Mycobacterium tuberculosis depends critically on the interaction and cooperation between monocyte-macrophages and T-lymphocytes and their cytokines
(Rook and Hernandez-Pando 1996). Substantial experimental evidence suggests that malnutrition can lead to secondary immunodeficiency that increase the host’s
16
susceptibility to infection. Severe malnutrition has profound effects on cell-mediated
immunity (Chandra 1991), and both macro- and micronutrient deficiency can influence
susceptibility to tuberculosis in animal models (Bhuyan and Ramalingaswami 1973;
McMurray et al. 1990). Increased risk of tuberculosis can result from alteration in the
individual protective function of, or the interaction between T-lymphocytes and macrophages because of nutritional insult (Chan J et al. 1997). Although a causal relationship has yet to be clearly established in humans, malnutrition may impair host responses in patients with tuberculosis, particularly leading to a vicious cycle of malnutrition and infection (Bhaskaram 1992).
An already malnourished individual with compromised cell mediated immunity is more likely to become infected with tuberculosis, and latent infection is more likely to become active tuberculosis. Among individuals with latent tuberculosis, the occurrences of malnutrition may be an important trigger for active tuberculosis development (Cegielski and McMurray 2004). One longitudinal study of participants in a BCG vaccine trial in the
United States found the incidence of active tuberculosis was 2.2 times higher in children with low subcutaneous fat stores (skin-fold thicknesses between 0 and 4 mm) compared to with those with 10 mm subcutaneous fat (Comstock and Palmer 1966). In a large study in Norway, the incidence of smear-positive and smear-negative tuberculosis declined significantly with increasing BMI in all age groups. New tuberculosis diagnosis was 5 times higher in the lowest BMI group compared with the highest BMI group (Tverdal
1986). Individuals with immuno-suppression have a greater risk of developing clinical
17
tuberculosis which explains the increased prevalence of tuberculosis in association with
HIV infection.
Effect of tuberculosis on nutritional status
During active tuberculosis, catabolic processes that cause wasting usually begins before the patient is diagnosed; therefore more is known about nutritional status at the time of diagnosis than of the wasting process per se (Macallan 1999). Prior studies have examined the effect of tuberculosis on nutritional state and demonstrated extensive nutritional depletion at the time of diagnosis (Onwubalili 1988; Zacharia R et al. 2002;
Zachariah et al. 2002). In developing countries, the impact of tuberculosis is even more dramatic. For example, one study in Malawi that evaluated 122 patients revealed a reduction in BMI of 20% from 21.7 to 17.3 with 35% reduction in skinfold thickness and
19% reduction in arm muscle circumference (Harries et al. 1988). Similar degrees of wasting have been reported in other populations (Scalcini et al. 1991). Such losses of both fat and fat-free mass (muscle or lean tissue) represent severe wasting and may pose significant impact on morbidity and mortality of their own in addition to that due to tuberculosis disease itself.
Effect of co-infection with HIV and tuberculosis on nutritional status
Although co-infection with HIV and tuberculosis may be thought to introduce an extra dimension to the pathophysiology of wasting, exacerbating the wasting seen in
18
tuberculosis or HIV infection alone (Macallan 1999; Lucas et al. 1994), recent studies
(Mupere et al. 2010) with a full panel of HIV positive and HIV negative adults with or without tuberculosis suggest gender but HIV as the factor associated with differences in body composition during co-infection. Further, tuberculosis has been suggested as the
dominant factor driving the wasting process at the time of diagnosis (Mupere et al. 2010;
Paton and Ng 2006). Prior studies that reported conflicting results were limited in terms
of sample size, comparison groups, and study populations that were composed of only
men (Paton et al. 1999) (Scalcini et al. 1991).
Possible Mechanism of Wasting or effect of tuberculosis on nutrition
In the 21st century, tuberculosis is still the most frequent underlying cause of wasting
worldwide. Wasting is regarded as the cardinal features of the disease. However, the
pathophysiology of wasting in tuberculosis remains poorly understood (Schwenk A and
Macallan D.C 2000). For any infection, there is a complex interaction between the host
response and the virulence of the organisms, which modulates the overall metabolic
response and the degree and the pattern of tissue loss. In patients with tuberculosis, a
reduction in appetite with eventual reduction in energy intake, nutrient malabsorption,
micronutrient malabsoption, and altered metabolism leads to wasting (Paton et al. 1999;
Macallan et al. 1998). However, factors that make the most dominant contribution in
tuberculosis have been difficult to identify because the wasting phase is rarely observed
and because of the ethical obligation to commence treatment once a diagnosis of
tuberculosis has been made. Observations have been made at the time of diagnosis that
19
the metabolic rate or resting energy expenditure is increased (Macallan et al. 1998).
However, by analogy with HIV infection where wasting phase has been scrutinized
(Macallan, Noble et al. 1995), increased energy expenditure is unlikely to be the sole driving force behind wasting in the catabolic phase. It is reduced energy intake which may be more likely to be the primary driving force.
In one study in India, patients with pulmonary tuberculosis were compared with malnourished and normally nourished healthy subjects. Whereas protein synthesis
(anabolism) and breakdown (catabolism) in the fasting state were not significantly different between groups, patients with tuberculosis used a larger proportion of proteins
(amino acids) from oral feeding for oxidation and hence for energy production than did either control group. The failure to channel food protein into endogenous protein synthesis has been termed “Anabolic block”. This anabolic block represents one of the mechanism for wasting in tuberculosis and other inflammatory status (Macallan et al.
1998; Paton et al. 2003). Tuberculosis may have a greater catabolic effect than HIV infection where similar studies failed to demonstrate anabolic block at the whole body level (Macallan, McNurlan et al. 1995). The difference in metabolic response may be the consequence of differences in the pattern of cytokine activation between tuberculosis and
HIV disease states.
Anorexia is also a contributing factor for wasting in tuberculosis. In an unselected U.S. cohort of patients diagnosed with tuberculosis, 45% lost weight and 20% had anorexia
20
(Miller et al. 2000). Increased production of cytokines with lipolytic and proteolytic
activity cause increased energy expenditure in tuberculosis (Verbon et al. 1999). Leptin
may also play an important role in wasting (Sarraf et al. 1997). In a study, malnutrition
has been associated with atypical presentations of tuberculosis (Madebo, Nysaeter, and
Lindtjorn 1997).
Micronutrient malnutrition in tuberculosis
Several micronutrient deficiencies have been described in individuals with tuberculosis
(van Lettow, Fawzi, and Semba 2003) and in those with HIV infection (Semba and Tang
1999; Bogden et al. 1990; Beach et al. 1992; Ullrich et al. 1994; Baum et al. 1991;
Ehrenpreis et al. 1994; Harriman et al. 1989). Several cross-sectional studies suggest that
patients with tuberculosis suffer from deficiencies of vitamin A (Evans and Attock 1971;
Smurova and Prokop'ev 1969), thiamin (Arkhipova O.P 1975), vitamin B6 (Miansikov
V.G 1969),folate (Markkanen et al. 1967; Line et al. 1971), and vitamin E (Panasyuk A.V
et al. 1991). Deficiencies that have been reported to be more prevalent among HIV-
infected adults than in those without HIV infection include vitamin A, vitamin E,
thiamin, riboflavin, vitamin B6, and vitamin C (Beach et al. 1992; Ullrich et al. 1994;
Baum et al. 1991; Ehrenpreis et al. 1994; Harriman et al. 1989). Of these deficiencies,
vitamin A and D have received the most attention in patients with tuberculosis. The
interest in vitamins A and D hinges on the historic use of cod-liver oil as treatment for tuberculosis prior to the era of antibiotics (Whalen C and Semba R.D 2001). Vitamins A,
C, E, B6, and folic acid and minerals zinc, copper, selenium, and iron all have key roles in
21
metabolic pathways, cellular function, and immune competence (Keusch 1990). The concentration of these may have a role in host defense against tuberculosis (Karyadi et al.
2000). Deficiency of single or multiple nutrients can reduce an individual’s resistance to any infection (Chandra 1991; Keusch 1990; Chandra and Kumari 1994).
Changes in Nutritional Status during Tuberculous Chemotherapy
During drug treatment of active tuberculosis without supplementary nutrition, nutritional status usually improves. This can be attributed to a variety of reasons including improved appetite and food intake, reduced energy/nutrient demands, and improved metabolic efficiency. Most improvements, how, are limited to increase in fat mass (Macallan 1999).
For example, a study by Schwenk et al. (Schwenk et al. 2004) that investigated the changes in fat mass and protein mass (fat-free mass) in 40 patients receiving standard tuberculosis treatment found after six months of treatment, the patients had gained 9.5 ±
8.95% body weight, mainly due to gain in fat mass with no significant change in protein mass suggesting that clinical recovery from tuberculosis does not guarantee protein mass restoration, even though weight gain is significant. This finding may support the idea that protein metabolism continues to be altered even during treatment, and that clinical and functional recovery from tuberculosis lags behind microbial cure. Alternatively, diet during treatment may have been inadequate in relationship to increased requirements during treatment and recovery, thereby limiting development of lean body mass.
22
Nutritional treatment of tuberculosis
Nutritional supplementation may help to improve outcome in tuberculosis patients. One study in Singapore (Paton et al. 2004), found that nutritional counseling to increase energy intake combined with provision of supplements, when started during the initial phase of tuberculosis treatment, produced a significant increase in body weight, total lean mass, and physical function after six weeks. A large proportion (46%) of the early weight gain comprised lean tissue, confirming the findings that tuberculosis can mount a protein anabolic response on feeding. In the same study, patients in the nutritional supplementation group continued to show a greater increase in body weight than control subjects during later follow-up. However, the pattern changed toward deposition of predominantly fat mass, whereas in the control group, the weight gain comprised fat lean tissue in approximately equal proportions (Paton et al. 2004). However, the changes in lean tissue described above could be an underestimate of the actual improvement in nutritional status, given that feeding initially leads to a loss of extracellular water that accumulates in malnourished individuals, including those with tuberculosis (Paton et al.
1999). Accelerating the recovery of lean tissue might help to restore physical functions more rapidly. Restoration of physical function might help to shorten the convalescent period and facilitate earlier return to productive work (Paton et al. 2004). Early restoration of nutrition could also lead to immunologic changes that could enhance the clearance of mycobacteria and reduce infectiousness of patient.
23
Vitamins and minerals can play important role in treatment of tuberculosis. In a trial among 110 new cases of active tuberculosis, subjects received tuberculosis chemotherapy alone, or in addition to injectable thiamin, vitamin B6, and vitamin C, or oral multivitamin supplement (Volosevich 1982). All groups receiving any vitamin supplementation had significantly better lymphocyte proliferation responses than the group receiving no supplement. Another trial showed that vitamins C and E were effective in improving immune responses to tuberculosis when given as adjuvant to multidrug tuberculosis therapy (Safarian et al. 1990). The supplementation with vitamin
A and zinc improved the effectiveness of the antituberculosis drugs in the first two months. The improved outcome was indicated by the higher number of patients with sputum negative for bacilli and significantly lower mean lesion area in the lungs (Karyadi et al. 2002).
24
CHAPTER 3
STUDY METHODS
32
Study Design
To address the aims of this dissertation, a hybrid of retrospective cohort and cross- sectional designs were employed. The completed five year Household Contact (HHC) study, the completed phase II prednisolone double blind randomized placebo controlled clinical trial (PD), and the ongoing Kawempe Community Health (KCH) study databases were used for the retrospective design. Datasets with relevant key variables from the
HHC, KCH, and PD were created and merged to form one large working database. A total of 753 patients were evaluated for analysis. Of the 753, 314 were enrolled into the
HHC, 344 into the KCH, and 95 into the placebo arm of the prednisolone clinical trial
(Figure 3:1). The datasets for the three studies were first tested for differences in baseline characteristics before combining for analysis. The datasets were different regarding extent of tuberculosis disease on chest x-ray because the phase II prednisolone trial enrolled only HIV-associated tuberculosis patients with CD4 cell count >200 cells/l compared to HHC and KCH studies (Appendix, Table 13:1). One of our interests was to establish the confounding effect of HIV; therefore, we combined all the three datasets for analysis. The BIA data (specifically, fat-free mass and fat mass) were collected during the KCH study only.
The HHC and KCH studies were observational epidemiologic studies; organized and conducted by the Makerere University and Case Western Reserve University TB research collaboration (Uganda-CWRU) that has been ongoing for the last 20 years in Uganda.
33
The HHC was the initial household contact study from 1995 to 1999 that described the epidemiology of TB in urban Kampala, Uganda (Guwatudde et al. 2003). The KCH is the second phase of the HHC. The KCH phase started in 2002 and is still ongoing (Stein et al. 2005). The KCH was developed specifically to focus on the determinants of host factors associated with primary infection, re-infection, reactivation, and progression of clinical disease and to identify and track individual strains of mycobacterial TB through
Ugandan households and local community. The phase II clinical trial was conducted between 1995 to 2000 to determine whether immunoadjuvant prednisolone therapy in
HIV-infected patients with TB who had CD4(+) T cell counts >/=200 cells/ mu L was safe and effective at increasing CD4(+) T cell counts (Mayanja-Kizza et al. 2005).
Patients were eligible for analysis if they were 18 or more years of age, had baseline measurements, had an HIV test, and were part of one prospective study conducted by the
Uganda-CWRU research collaboration. Adults with a previous history of treated pulmonary TB were excluded in the study.
In a cross-sectional study, 132 participants 18 years or older residing in Kampala district or 20 km from the study site if residence was outside Kampala in Uganda were enrolled.
One participant was excluded from the analysis because of prior TB treatment. The study was conducted at the National TB and Leprosy Program (NTLP) Clinic of the national tertiary teaching hospital complex, Mulago between November 2007 and March 2008. Of the 131 participants who were included in the analysis, 31 were HIV positive with TB and 32 were HIV negative with TB and were recruited at the Mulago NTLP Clinic; 38 were HIV positive without TB and recruited at the Infectious Disease Institute Clinic
34
(IDI) located 500 meters from the Mulago NTLP Clinic; and 30 HIV negative individuals without TB were enrolled from the community where TB patients resided (Figure 3:2).
The retrospective cohort and cross-sectional designs were chosen in view of their advantages. There was a rich existing database on study participants who fit the needed criteria for a retrospective cohort design. A cross-sectional design could easily be implemented in a short period of time for primary data collection to study associations.
The retrospective cohort and cross-sectional were thus cost effective and took time constraints into consideration.
In a retrospective cohort, the investigator identifies a cohort of individuals based on their characteristics in the past and then reconstructs their past or up to the present time (or occasionally into the future). The ideal concurrent cohort design in which the investigator identifies the cohort on the basis of current characteristics and is then followed forward in time would be expensive and would require a longer time to complete the study. The potential bias of missingness in data anticipated in a retrospective cohort design was assessed for using standard statistical procedures.
In a cross-sectional design, the investigator evaluates for the exposure and outcome of interest at the same one point in time. The disadvantage with a cross-sectional design is that the presence or absence of both exposure and disease/outcome of interest are
35
determined at the same time in each individual in the study. It is not possible to establish
a temporal relationship between exposure and onset of disease. Therefore, although a
cross-sectional study can be very suggestive of a possible risk factor or factors for a
disease, with limitations in establishing the temporal relationships, the association cannot
reflect a causal relationship.
The institutional review boards at Case Western Reserve University in the United States
and Joint Clinical Research Center in Uganda reviewed the protocol and final approval
was obtained from the Uganda National Council for Science and Technology. All
patients had written informed consent taken to be enrolled in the parent studies. All
participants were given appropriate pre- and post-test HIV counseling and AIDS education.
Measurements
In all the four studies, socio-demographic and clinical information was obtained through standardized interviews and physical examination performed by trained medical officers.
Venous blood was collected for HIV-1 enzyme immunoassay testing and complete blood and differential counts. HIV infection was documented by enzyme-linked immunosorbent assays. None of the HIV positive patients, neither those who were newly identified with
HIV nor those with pre-existing HIV, were on antiretroviral therapy. All participants had posterior-anterior chest X-rays taken at baseline. Expectorated sputum specimens were collected, concentrated, and stained for acid fast bacilli (AFB) with Ziehl-Neelsen stain at
36
the Wandegeya national reference laboratory in Uganda. AFB smears were reviewed by
trained technicians who graded the smears by the number of acid-fast organisms seen on
the light microscopy according to criteria established by the WHO (International Union
Against Tuberculosis and Lung Disease 1986). Specimens were cultured for
mycobacteria tuberculosis on Lowenstein-Jensen medium slants, incubated at 370C in air and examined weekly until positive or for 8 weeks. Patients with active tuberculosis were treated with standard four-drug chemotherapy for tuberculosis per guidelines of the
Ugandan Ministry of Health.
Nutritional status was assessed using anthropometric measurements such as height and
weight and BIA Detroit, MI, RJL Systems. Body weight was determined to the nearest
0.1 kg using a SECA adult balance, and standing height was determined to the nearest 2 mm. Body-mass index (BMI) was computed using the relationship of weight in kilograms divided by height in meters squared (kg/m2). All BIA measurements were
performed by one trained observer (except measurement for the cross-sectional study that
had two medical officers) using the same equipment and recommended standard
conditions with regard to body position, previous exercise, dietary intake, skin
temperature, and voiding of the bladder were taken into consideration in taking BIA
measurements (Kyle et al. 2004). All BIA measurements during the KCH study were
performed on the day patients were confirmed to have TB disease.
37
The BIA is a simple, easy, safe, non-invasive technique, that has been recommended for nutritional studies in the clinical setting (Kyle et al. 2004; Kyle, Piccoli, and Pichard
2003) and is a convenient method to determine the lean or fat-free mass and fat body compartments (Kyle, Piccoli, and Pichard 2003; Kyle et al. 2004). Single-frequency BIA was performed at 50 kHz and 800 mA with standard tetrapolar lead placement (Jackson et al. 1988). Before performing measurements on each participant, the BIA instrument was calibrated using the manufacturer’s recalibration device. The resistance and reactance were based on measures of a series circuit (Kotler et al. 1996). BIA measurements were performed in triplicate for each subject. Fat-free mass was calculated from BIA measurements using equations that were previously cross-validated in a sample of patients (white, black and Hispanic) with and without HIV infection (Kotler et al.
1996) and have been applied elsewhere in African studies (Villamor et al. 2006; Shah et al. 2001; Van Lettow et al. 2004). Fat mass was calculated as body weight minus fat-free
mass.
Operational definition
We used BMI and height-normalized indices (adjusted for height2) of body composition
that partition BMI into fat-free mass index (FFMI) and fat mass index (FMI) (Schutz,
Kyle, and Pichard 2002; VanItallie et al. 1990; Kyle, Piccoli, and Pichard 2003) to
establish the body wasting status of participants. The FFMI and FMI have the advantages
of compensating for differences in height and age (Kyle, Genton, and Pichard 2002).
Also, the use of the FFMI and FMI eliminates some of the concerns about differences
38
between population groups. We defined body wasting as patients having the low fat-free
mass index (FFMI) and the low body fat mass index (FMI) corresponding to WHO BMI
categories for malnutrition as previously reported (Kyle, Piccoli, and Pichard 2003). The
FFMI <16.7 (kg/m2) for men and <14.6 (kg/m2) for women and the FMI <1.8 (kg/m2) for
men and <3.9 (kg/m2) for women corresponds to a BMI of <18.5 kg/m2, the WHO cutoff
for malnutrition (World Health Organ Tech Rep 1995) among adults.
Statistical Analyses
Descriptive statistics
This involved distribution of all variables of interest by computing frequencies and
proportions for categorical variables, means and medians for continuous variables. In
measures analysis, plots of individual participants’ and group mean trajectory response
profiles were performed to assess how BMI, FFMI, and FMI evolved over time and how
measurements made at different time points were related.
Bivariate associations
Chi-square and Fisher’s exact tests were used to assess associations and differences in
proportions between categorical variables. Fisher’s exact test was used where tabular
counts were less than five. Student’s t-test and analysis of variance (ANOVA) were used to compare continuous variables across categories. Variance ratio testing was performed to determine whether to use equal or unequal variances for the Student’s t-test and
39
Bartlett’s test for equal variances was used to assess equality of variance assumption for
ANOVA. Non-parametric test (Mann-Whitney test) was used where continuous variables were not normally distributed or when sample size distribution was small.
Stratified analysis
Stratified analysis was performed according to key confounding variables such as gender
and HIV to assess for confounding and effect modification.
Univariate and multivariable analyses
Univariate analyses were performed to assess unadjusted associations between the
outcome variables and the main predictor and the potential confounders.
Multivariable analyses were performed using multivariable modeling to estimate adjusted
measures of associations while taking into account multiple confounding variables.
Variables for inclusion in the multivariable models were selected based on biological
plausibility and statistical significance in univariate analysis.
Further analyses were performed according to each specific aim.
40
Figure 3:1 Retrospective cohort study flow
658 Household Contact 95 Placebo arm Prednisolone
753 Study participants
Exclusion: 6 <18 yrs
Subset with BIA: 311
BMI: wasted No wasting FFMI: Wasted No wasting 310 (42%) 437 (58%) 103 (33%) 208 (67%)
Repeated Measurements: Baseline & Follow-up: BMI at 0, 2, 3, 5, 6, 12, and 24 FFM, FM at 0, 3, 12, and 24
Household contact includes 344 from the initial household contact study and 314 from the Kawempe Community Healthy study. BMI = body mass index, FFM = fat-free mass, FFMI = fat-free mass index, FM = fat mass, and FMI = fat mass index
41
Figure 3:2 Cross-sectional study flow diagram
137 Screened >18 years Exclusions: 4 Declined 1 Appointment 1 Prior TB Rx 131 included in analysis
32 TB no HIV 30 TB with HIV 38 HIV no TB 31 no HIV no TB
Measurements: 24-Hour Dietary Intake Recall Weight, height, BMI, Fat-free mass, Fat mass Clinical severity of TB disease: TBscore with 13 clinical variables; 0-5 mild and >5 moderate/severe disease
TB = tuberculosis, BMI = body mass index.
42
CHAPTER 4
BODY COMPOSITION MEASURED WITH BIOELECTRICAL IMPEDANCE
ANALYSIS AND ANTHROPOMETRY AMONG HIV POSITIVE AND HIV
NEGATIVE ADULTS WITH OR WITHOUT TUBERCULOSIS IN KAMPALA,
UGANDA
46
Abstract
Background Simple and inexpensive method for assessing body composition is lacking in sub-Saharan Africa, a region with high burden of tuberculosis and HIV dual epidemic to ensure early detection and timely management of population groups at risk of fat-free
mass (FFM, lean tissue) wasting associated with poor outcome.
Objective We determined to establish whether existing equations that involve simple
anthropometric measurements of waist circumference (WC), body mass index (BMI), or
mid-upper arm circumference (MUAC) to provide comparable results to those of
bioelectrical impedance analysis (BIA).
Method In a cross-sectional study of 131 participants who were screened for active
tuberculosis and HIV infection, reactance, resistance, height, weight, MUAC, waist and
hip circumferences, and four skinfold thickness (SF): triceps, biceps, subscapular, and
suprailiac were measured. We compared fat mass and FFM as measured by BIA and as
measured by equations that involved waist circumference, BMI, or MUAC using Bland-
Altman plots.
Results On average, the equation that involved WC showed no bias in estimating
comparable FFM to that of BIA among men with tuberculosis regardless of HIV status
while in women, the equation showed no bias among individuals without tuberculosis
regardless of HIV status. The equation showed no bias in estimation of fat mass in
comparison to BIA among men and women without tuberculosis regardless of HIV
status. The equation that involved four SF underestimated FFM and overestimated fat
mass among men and among women regardless of tuberculosis and HIV status. There
47
was no bias among men and women with tuberculosis regardless of HIV status as regards the equation that involved BMI in estimation of FFM. Concerning fat mass determined by this equation, there was no bias among women without tuberculosis and there was no bias in estimating fat mass among men with tuberculosis regardless of HIV status. There was no bias in fat mass and FFM as determined by the equation that involved MUAC in comparison to BIA among men and women regardless of tuberculosis and HIV status.
Conclusion The existing equations with simple anthropometric measurements that provided comparable results to BIA differed by gender and tuberculosis disease status.
During active tuberculosis, equations that involved MUAC and BMI providing comparable results of body composition to those of BIA among men and among women whereas in the absence of active tuberculosis among men, the equation with MUAC provided comparable results and among women, the equation with WC provided comparable results regardless of HIV status. Further studies are needed to validate the
BIA and the equation with simple anthropometric measurements in assessing body composition in an African population.
48
Background
Body wasting and malnutrition are associated with tuberculosis, and co-infection with
HIV and tuberculosis may be potentially exacerbate the wasting that occurs in
tuberculosis or HIV infection alone (Macallan 1999; Lucas et al. 1994; Schwenk A and
Macallan D.C 2000; van Lettow, Fawzi, and Semba 2003). In sub-Saharan Africa with
the highest overlapping tuberculosis and HIV infection epidemic, by the time patients
present for registration and treatment, a significant proportion have a profound degree of
body wasting and malnutrition (Niyongabo et al. 1999; Kennedy et al. 1996; Zachariah et
al. 2002). Moreover, progressive loss of body mass in TB and HIV-related wasting are strong risk factors for morbidity, mortality, and impaired physical function (Zachariah et al. 2002; Van Lettow et al. 2004; Tang et al. 2002; Harries et al. 1988). Body weight
compartments of fat and fat-free mass differ in their contribution to body wasting and in
their contribution to clinical benefit. For example, fat-free mass (the lean tissue) is more
closely correlated with morbidity, mortality, quality of life, and physical functioning than
fat mass and body weight itself (Wagner, Ferrando, and Rabkin 2000; Mostert et al.
2000; Heitmann et al. 2000).
Assessment of body composition in tuberculosis and HIV patients in sub-Saharan Africa,
a region with 79% of the overlapping global HIV-tuberculosis disease burden, (Lawn and
Churchyard 2009) is critical to ensure prevention, early detection, and timely management of body wasting. Further, assessment is crucial to understand the
49
pathophysiology and body compartments involved in body wasting. However, most of
the available reference laboratory-based techniques (Norgan 2005), such as air-
displacement plethysmography (ADP), underwater weighing and dual X-ray absorptiometry (DXA) that provide accurate data require a high level of technical expertise; they are time-consuming, unportable, and expensive for field settings in sub-
Saharan Africa. Bioelectrical impedance analysis (BIA) is an easy, safe, non-invasive,
and convenient method recommended to determine to fat and fat-free mass body
compartments (Kyle, Piccoli, and Pichard 2003; Kyle, Bosaeus, De Lorenzo, Deurenberg,
Elia, Gomez et al. 2004); however, its use in sub-Saharan Africa is still limited to
research settings and validation studies are lacking. Skinfold thickness (SF) measurement
is a simple and inexpensive recommended method (Norgan 2005); however, it requires
multiple measurements and the most widely used Durnin and Womersley equation
(Durnin and Womersley 1974) which was established in a Caucasian sample has been
found to be unsuitable in African populations (Oosthuizen et al. 1997).
To fill the current gap in assessing body composition in routine clinical and
epidemiological settings, and to validate population specific equations for use in sub-
Saharan Africa, we present results showing simple existing equations that involve waist
circumference, body-mass index (BMI), and mid-upper-arm circumference (MUAC) to
provide comparable fat and fat-free mass body composition to that of BIA among HIV
positive and HIV negative adults with/or without tuberculosis in urban Kampala, Uganda.
50
Methods
Subjects
In a cross-sectional study, 132 participants 18 years or older residing in Kampala district or 20 km from the study site if residence was outside Kampala in Uganda were enrolled.
One participant was excluded from the analysis because of prior tuberculosis treatment.
The study was conducted at the National tuberculosis and Leprosy Program (NTLP)
Clinic of the national tertiary teaching hospital complex, Mulago between November
2007 and March 2008. Of the 131 participants who were included in the analysis, 31
were HIV positive with tuberculosis and 32 were HIV negative with tuberculosis and
were recruited at the Mulago NTLP Clinic; 38 were HIV positive without tuberculosis
and recruited at the Infectious Disease Institute Clinic (IDI) located 500 meters from the
Mulago NTLP Clinic; and 30 HIV negative individuals without tuberculosis were
enrolled from the community where TB patients resided. The institutional review boards
at Case Western Reserve University and Joint Clinical Research Center approved the
study, with final approval by the Uganda National Council for Science and Technology.
All participants provided written informed consent to the study.
All subjects in the study were given appropriate pre- and post-test HIV counseling and
AIDS education. HIV-1 infection was diagnosed on the basis of a positive enzyme-linked immunosorbent assay for HIV-1 antibodies (Recombigen; Cambridge Biotech,
Cambridge, MA). At enrollment, basic demographic information and a medical history
51
were collected, and a standardized physical examination was conducted by a medical
officer. Active pulmonary tuberculosis was confirmed by sputum smear microscopy and
culture. Patients with active tuberculosis were treated with standard four-drug chemotherapy for tuberculosis per guidelines of the Ugandan Ministry of Health.
Similarly, HIV positive patients who were eligible for antiretroviral therapy were started on treatment and cotrimoxazole prophylaxis at the IDI clinic.
Body composition assessment
From anthropometry
Anthropometric measurements included height, weight, MUAC, waist and hip circumferences, and four SF: triceps, biceps, subscapular, and suprailiac. BMI was computed using the relationship of weight in kilograms divided by height in meters squared (kg/m2). Weight was taken using Hanson digital electronic scales to the nearest
100g. Standing height was measured to the nearest cm using a stadiometer. All
circumferences were measured to the nearest 0.1 cm with a non-elastic metric measuring
tape. The left MUAC was measured over the biceps at the mid-point between acromion
and olecranon with the participant’s arm relaxed. The waist circumference was measured
on a horizontal plane at the narrowest part of the torso, i.e., the smallest horizontal
circumference in the area between the ribs and iliac crest. For measurement of the hip
circumference, the measurer squatted beside the participant to judge the level of
maximum extension of the buttocks. The circumference was measured at this level on a
horizontal plane. The SF measurements were taken on the left side of the body using a
52
Lange skinfold caliper model 12-1110 to nearest millimeter in each participant. All
reported anthropometric measurement values were the mean of duplicates.
The measurements were made using standard procedures (Lohman T.G, Roche A.F, and
Martorell R 1988; Frisancho A.R 1990) by two trained medical officers who took the
standard history and physical examinations. Technical errors of measurements (TEM)
were computed for each trained officer by measuring height, weight, MUAC, waist and
hip circumferences, and four SF of a set of 6 individuals. The trained officer and the
supervisor made measurements of each individual in succession (reading 1). Serial
second measurements of MUAC were then obtained (reading 2). Intra-subject reliability was tested by performing using the duplicates. Intra-subject differences were calculated in absolute values. The TEM =
(reading 1 reading 2)squared/2 X number of duplicates) and the percent
�reliability∑ (TEM− x 100/overall mean of the measurements) for all measurements of each
trained medical officer were calculated. Based on six repetitions, the TEM and reliability
coefficient were within acceptable limits (Lohman T.G, Roche A.F, and Martorell R
1988; Frisancho A.R 1990) (Appendix, Tables 12:1, 12:2, and 12:3).
The following existing equations validated among Caucasians by Lean et al. (Lean, Han,
and Deurenberg 1996) were used to estimate percentage fat mass. The equations were
developed from different combination of anthropometric measurements including waist
circumference, BMI, and MUAC adjusted for sex and age. We selected these equations
53
because few simple measurements are needed to estimate fat mass compared to using SF.
Furthermore, they were shown to have good prediction of body density with least bias,
such as the one involving waist circumference in both men and women. Fat-free mass was computed by subtracting fat mass from weight in kilograms.
Percent fat mass = (0.567 x waist circumference) + (0.101 x age) – 31.8 for men
= (0.439 x waist circumference) + (0.221 x age) – 9.4 for women
= (1.33 x BMI) + (0.236 x age) – 20.2 for men
= (1.21 x BMI) + (0.262 x age) – 6.7 for women
= (1.52 x MUAC) + (0.336 x age) – 38.7 for men
= (1.38 x MUAC) + (0.243 x age) – 16.7 for women.
In addition, the widely used Durnin and Womersely equation (Durnin and Womersley
1974) that transforms the log sum of the four SF was used to show its lack of agreement with the BIA body composition.
Body density = 1.1765 – 0744 x log∑(triceps + biceps + subscapular + suprailiac) for men,
= 1.1567 – 0.0717 x log∑(triceps + biceps + subscapular + suprailiac) for women.
Body density is converted into percent body fat using Siri’s equation (Siri W.E 1961).
Percent fat mass = [(4.95/body density) – 4.5] x 100.
54
From bioelectrical impedance analysis
The single-frequency bioelectrical impedance analyzer (BIA Detroit, MI, RJL Systems) performing at 50 kHz and 800 mA was used for BIA measures with detecting electrodes placed on the wrist and ankle and signal introduction electrodes placed on the first joint of the middle finger and behind the middle toe. Before performing measurements on each subject, the BIA instrument was calibrated using the manufacturer’s recalibration device.
The resistance and reactance were based on measures of a series circuit (Kotler et al.
1996). BIA measurements were performed in duplicate for each subject. The analyzer was calibrated monthly. Fat-free mass was calculated from BIA measurements using equations that were previously cross-validated in a sample of patients (white, black and
Hispanic) with and without HIV infection (Kotler et al. 1996) and have been applied elsewhere in African studies (Shah et al. 2001; Van Lettow et al. 2004; Villamor et al.
2006). Fat mass was calculated as body weight minus fat-free mass.
Analysis
The characteristics of men and women were compared using Wilcoxon-Mann-Whitney test for continuously distributed variables due to lack of normality. Height-normalized indices (adjusted for height2) of fat mass (FMI) and fat-free mass (FFMI) were computed
(Schutz, Kyle, and Pichard 2002; VanItallie et al. 1990; Kyle, Piccoli, and Pichard 2003) and compared between women and men. The FFMI and FMI have the advantages of compensating for differences in height and age (Kyle, Genton, and Pichard 2002). Also,
55
the use of the FFMI and FMI eliminates some of the differences between population
groups.
Fat mass and fat-free mass measured by BIA were compared with the corresponding
anthropometric measures using Spearman correlation coefficients (r) and the method of
Bland and Altman (Bland and Altman 1986), by plotting the bias (BIA measure –
anthropometric measure) against the average of the measure by BIA and anthropometric.
The average bias was calculated as the mean difference in the BIA and the
anthropometric measures and limits of agreement as mean difference ± 1.96 x S.D. To
evaluate whether bias was constant across all levels of the measure, the least-squares
regression slope of the bias was estimated. A slope difference from zero indicated that the
magnitude of bias depended on the level of the measure. Paired t-tests were used to detect
significant differences in body composition obtained using BIA compared to
anthropometric method. The main hypothesis tested was that there were no differences in
measures of fat mass or fat-free mass by BIA compared to anthropometric method (i.e.,
zero mean bias) using a two-tailed α of 0.05. Stratified analysis was performed according
to gender and HIV status. All analyses were performed using SAS software version 9.2
(SAS Institute, Cary, North Carolina).
Results
Overall men were taller, had more fat-free mass as measured by BIA and equation
involving waist circumference, and more fat mass as estimated using equation that
56
involve a sum of four SF. However, men had lower BMI, SF, hip circumference, and fat mass as measured by BIA and equation involving waist circumference compared to women regardless of tuberculosis and HIV status (Tables 4:1 and 4:2). Men and women had similar waist circumference and MUAC regardless of tuberculosis and HIV status.
There were strong positive correlation coefficients between the measurements made using BIA and the measurements made using equations that involved waist circumference, BMI, or MUAC as presented in Tables 4:3 to 4:7. The high correlation coefficients indicate strong relationships between the measurements made using BIA and the measurements made using equations that involved waist circumference, BMI, or
MUAC. However, a comparison of differences between the measurements made using
BIA and the measurements made using equations that involve anthropometric measures for individual subjects shows that the high correlation coefficients disguise large inter- method bias and error. For example in table 4:3, whereas BIA and the equation involving sum of four SF correlated well (r = 0.36, p<0.001) for estimation of fat mass, the equation involving sum of four SF overestimated fat mass by a mean of 19.41 with an error of 6.76.
Fat-free mass determined by the equation using waist circumference showed no bias in the total study population (Table 4:3, Appendix Figure 12:1) and no bias among all men
(-0.50 ± 3.53, p=0.27) and among all women (0.31 ± 3.94, p=.52; Table 4:4) compared to
BIA. On average, the equation showed no bias among men with TB regardless of HIV
57
status while in women, the equation showed no bias among individuals without TB regardless of HIV status. However, fat-free mass was overestimated among HIV negative men with no tuberculosis (-2.42 ± 2.96, p=0.005) and underestimated among HIV positive women with TB (2.45 ± 2.71, p=0.001), but in both the bias did not depend on the level of fat-free mass. The slopes of bias were 0.29 and -0.05, respectively with p- value >0.05 for both (Table 4:4). The equation showed no bias in estimation of fat mass in comparison to BIA among men and women without TB regardless of HIV status although on average there was overestimation of fat mass in the overall population (Table
4:3, Appendix Figure 12:2), in all men and in all women (Table 4:4).
The Durnin and Womersely equation that involved sum of four SF underestimated fat- free mass in the total study population (17.86 ± 7.38, p<0.001) and it overestimates fat mass (-19.41 ± 6.76, p<0.001) compared to BIA (Table 4:3, Appendix Figure 12:3). The bias in fat-free mass and in fat mass measures were depended on the level of fat-free mass and fat mass; the slopes of bias were 0.50 and 0.78, respectively with p<0.001. In general, the Durnin and Womersely equation underestimated fat-free mass and overestimate fat mass among men and among women regardless of TB and HIV status
(Table 4:5, Appendix 12:3 and 12:4).
The fat-free mass determined by the equation using BMI showed no bias in the overall population in comparison to BIA (Table 4:3, Appendix Figure 12:5). Specifically in stratified analysis, there was no bias among men and women with TB regardless of HIV
58
status (Table 5). Regarding fat mass determined by this equation, there was no bias among women without TB regardless of HIV status (Table 5, Appendix 12:6 fat mass for overall population). The slope (0.05, p>0.05) of the bias for fat mass among HIV positive men with TB was not significant suggesting that the bias in fat mass measure was not dependent on fat mass (Table 5). Thus, the equation had no bias in estimating fat mass in comparison to BIA among men with TB regardless of HIV status.
For the entire study population on average, the equation using MUAC overestimated fat- free mass in comparison to BIA (-1.40 ± 3.39, p<0.001), but the bias in the fat-free measure was not dependent on fat-free mass (Table 2, Appendix Figure 12:7). Similarly in stratified analysis regardless of gender and HIV status, the slopes of the bias for fat- free mass were not significant suggesting lack of dependence for the bias in fat-free mass measure on the fat-free mass (Table 6). One can therefore say that there was no bias in fat-free mass determined by the equation involving MUAC in comparison to BIA among men and women regardless of TB and HIV status. Following the same argument, among men regardless of TB and HIV status, the equation showed no bias in fat mass estimation compared to BIA (Table 6, Appendix Figure 12:8). The equation also showed minimal bias in estimating fat mass among women regardless of TB and HIV status.
Discussion
This is the first cross-sectional study in sub-Saharan Africa to show equations with simple anthropometric measurements that provide comparable results of body
59
composition among adults to that of BIA. In the present cross-sectional study of 131 participants, we tested existing equations (Lean, Han, and Deurenberg 1996) that use waist circumference, BMI, or MUAC to provide comparable results of body fat and fat- free mass with BIA in a population of HIV positive and HIV negative adults with/or without active TB. The equations that provided comparable results to BIA differed by gender and TB disease status. The equation that involved MUAC or BMI provided comparable results of fat-free mass to that of BIA whereas the equation that involved
MUAC provided comparable results of fat mass among men and among women with TB regardless of HIV status. Among men without TB, the equation that involved MUAC provided comparable results of fat and fat-free mass to that of BIA whereas among women without TB, the equation that involved waist circumference provided comparable results of fat and fat-free mass regardless of HIV status. The equations involved a sum of four SF by Durnin and Womersely (Durnin and Womersley 1974), provide biased estimates of fat and fat-free mass that were not comparable to BIA regardless of gender and TB status.
The findings in this study appear to suggest that active TB, not HIV infection, may be the determining factor to consider in selecting equations that will provide comparable results of fat and fat-free mass to that of BIA. In the face of active TB, equations that involved
MUAC and BMI were the equations that had no or minimal bias in providing comparable results of body composition to those of BIA among men and among women regardless of
HIV status. In the absence of active TB among men, the equation that involved MUAC had no or minimal bias in providing comparable results of body composition to those of
60
BIA whereas among women, the equation that involved waist circumference had no or minimal bias in providing comparable results regardless of HIV status. To our knowledge, this is the first study in sub-Saharan Africa to provide evidence for selecting appropriate equations that involve simple anthropometric measurements for use in estimating fat and fat-free mass among men and women. Previous studies in Africa
(Oosthuizen et al. 1997; Dioum et al. 2005), have evaluated the Durnin and Womersley
(Durnin and Womersley 1974) equation that involve a sum of four SF with biased results of body composition similar to what has been revealed in the present study. The major strengths of our study stem from the heterogeneity of the study population that involved a full panel of HIV positive and HIV negative adults with/ or without active TB among men and women.
In the present study, equations that use simple anthropometric measurements to provide comparable results of body composition to those of BIA among men and women differed by TB disease status. The equations that involve MUAC or BMI provided comparable results of body composition to those of BIA among men and women with TB whereas the equation with MUAC and the equation with waist circumference provide comparable results among men and among women without TB, respectively regardless of HIV status.
The possible explanation to this gender difference in the face of active TB appears to reflect the effect of TB on the nutritional status of patients with the disease. Further, it reflects the dominant nature of TB disease in inducing the wasting process at the time of diagnosis among patients with HIV associated TB. This finding been has reported previously (Paton and Ng 2006; Mupere et al. 2010), and it is reflected in the results of
61
the present study. For example in this study, in the presence of active TB both men and women had low comparable BMI and MUAC whereas in the absence of active TB, women had BMI and MUAC of higher magnitude compared to men. Moreover, MUAC has been found to reflect adult nutritional status as defined by BMI (Collins 1996;
Powell-Tuck and Hennessy 2003), and MUAC is effective in the determination of malnutrition among adults in developing countries (Bisai and Bose 2009; James et al.
1994). Further, MUAC is a simpler measure than BMI, requiring a minimum of equipment and in practice has been found to predict morbidity and mortality as accurately as deficits in weights (Briend et al. 1989). The equation with waist circumference providing comparable results among women in the absence of active TB is in line with the known fact that women have a higher content of body fat compared to men (Blaak
2001). Moreover, waist circumference has been reported as sensitive indicator of total and central adiposity fat (Taylor et al. 1998; Flegal et al. 2009).
The interpretation of results in the present study is not without limitations. In this study, we used the BIA method in measurement of body composition, yet it is not of reference standard like the dual-energy x-ray absorptiometry. The BIA prediction method used has not yet been validated in the local population. As a result, findings of body composition may be biased because of variations in hydration across ethnic groups (Kyle, Bosaeus, De
Lorenzo, Deurenberg, Elia, Manuel Gomez et al. 2004). However, the equations that were used in this study were previously cross validated in individuals of different race
(white, black, and Hispanic) among men and women, who were both healthy controls and
HIV-infected patients (Kotler et al. 1996). Moreover, the equations have been used
62
widely in other studies from Africa with meaningful findings (Shah et al. 2001; Van
Lettow et al. 2004; Villamor et al. 2006; Mupere et al. 2010). Also care was taken in
taking measurements at rest, with proper placement of leads, in participants who had not
exercised or taken alcohol, in participants with voided bladder and ambient temperature.
However, half of the measurements were in patients with underlying illness that may
cause shifts in body water compartments, thereby affecting measurements of fat mass.
Our findings are also limited by the cross-sectional nature of the study that as the body composition change while TB patients receive treatment, selection of appropriate of
equations may change with time. We cannot evaluate such temporal changes in the
present study.
Despite limitations of the present study, findings in this study revealed remarkable gender
in the presence or absence of active TB but not HIV that may influence selection of
appropriate equations with waist circumference, BMI, or MUAC to provide comparable
results of body composition to that of BIA. Further studies are needed to validate the BIA
and the equation with simple anthropometric measurements in assessing body
composition in an African population.
Acknowledgements
We thank all study staff members of the Case Western Reserve University and Makerere
University research collaboration at the Tuberculosis Research Unit in the United States
and in Uganda for their assistance; the faculty of staff at Case Western Reserve
63
University Department of Epidemiology and Biostatistics for the guidance in analyzing the project; and the Fogarty International Center, for the continued support.
This study was supported in part by the AIDS International Training Research Program,
Fogarty International Center, Grant No. TW000011, based at Case Western Reserve
University, Department of Epidemiology and Biostatistics. This work was part of Ezekiel
Mupere’s PhD thesis at Case Western Reserve University.
64
Table 4:1 Characteristics of study population with tuberculosis
Characteristics (SD)1 HIV positive with TB HIV negative with TB
Men Men Women Women (n=10) (n=21) (n=18) (n=14) Age, years 30.9 (4.6) 29.2 (5.9) 26.0 (7.3) 26.0 (5.4) Weigh, kg 54.3 (6.1) 46.9 (7.4)b 53.2 (6.5) 49.7 (7.6) Height, cm 171.8 (9.3) 158.7 (6.8)a 171.1 (6.0) 157.7 (12.8)a BMI, kg/m2 18.4 (1.7) 18.6 (3.0) 18.2 (2.0) 20.3 (4.3) MUAC, mm 23.0 (2.2) 23.4 (3.0) 23.7 (1.8) 24.5 (2.5) Triceps ST, mm 6.4 (2.3) 13.2 (6.6)b 6.3 (2.0) 16.6 (5.6)a Biceps ST, mm 3.8 (1.0) 6.6 (3.7) 3.6 (0.9) 6.8 (2.7)a Subscapular ST, mm 8.3 (2.7) 11.7 (4.7)b 8.5 (1.9) 13.2 (5.9)b Sacral ST, mm 5.6 (1.7) 9.2 (5.1) 5.4 (1.6) 9.8 (5.1)a Waist circumference, cm 72.5 (4.7) 70.1 (6.3) 70.0 (3.7) 68.5 (5.2) Hip circumference, cm 84.7 (5.1) 86.5 (7.7) 84.4 (4.7) 90.9 (8.1)b BIA FFM, kg 48.7 (6.0) 36.0 (3.8)a 48.1 (5.5) 37.6 (6.2)a BIA FFMI, kg/m2 16.6 (1.3) 15.4 (0.9)b 16.6 (1.5) 16.1 (0.9) WC FFM, kg 47.5 (4.5) 33.6 (4.0)a 47.6 (5.0) 36.6 (4.7)a WC FFMI, kg/m2 16.1 (1.4) 13.3 (1.6)a 16.2 (1.5) 15.1 (3.2)b BMI FFM, kg 48.0 (4.9) 35.6 (4.1)a 47.7 (4.9) 37.3 (4.8)a BMI FFMI, kg/m2 16.3 (1.2) 14.1 (1.4)a 16.3 (1.3) 15.2 (2.1)b MUAC FFM, kg 50.6 (4.2) 35.9 (3.9)a 49.9 (5.2) 38.0 (4.3)a MUAC FFMI, kg/m2 17.2 (1.5) 14.2 (1.6)a 17.0 (1.4) 15.6 (3.0)b BIA Fat mass, kg 5.2 (1.8) 8.3 (5.2)a 4.8 (2.3) 10.5 (6.6)b BIA FMI, kg/m2 1.8 (0.6) 3.3 (2.1)a 1.6 (0.8) 4.6 (3.6)a WC Fat mass, kg 6.8 (2.1) 13.2 (3.5)a 5.7 (1.8) 13.3 (3.3)a
65
WC FMI, kg/m2 2.3 (0.6) 5.2 (1.5)a 1.9 (0.6) 5.4 (1.5)a BMI Fat mass, kg 6.3 (1.6) 11.2 (3.7)a 5.5 (2.3) 12.7 (4.1)a BMI FMI, kg/m2 2.2 (0.6) 4.5 (1.6)a 1.9 (0.8) 5.3 (2.4) a MUAC Fat mass, kg 3.7 (2.2) 10.9 (3.8)a 3.4 (2.3) 12.0 (3.7)a MUAC FMI, kg/m2 1.2 (0.7) 4.3 (1.5)a 1.2 (0.8) 4.9 (1.6)a ∑4 SF FFM, kg 23.3 (3.5) 21.2 (4.7)b 22.8 (3.6) 23.3 (4.9) ∑4 SF FFMI, kg/m2 7.9 (1.0) 8.4 (1.9) 7.8 (1.1) 9.6 (2.5)b ∑4 SF Fat mass, kg 31.0 (2.7) 25.7 (2.7) 30.4 (3.0) 26.7 (3.0)b ∑4 SF FMI, kg/m2 10.5 (0.8) 10.2 (1.1) 10.4 (1.0) 10.9 (2.0)
ªp-value < 0.001, bp-value < 0.05; p-values were obtained with Mann-Whitney test and were comparisons between men and women. 1Values are means ± standard deviation (SD). TB = Tuberculosis disease, WC = waist circumference, MUAC = mid-upper arm circumference, SF = Skinfold thickness, BMI, body mass index, FFM = Fat-free mass, FFMI = Fat-free mass index, and FMI = Fat mass index.
66
Table 4:2 Characteristics of study population without tuberculosis
HIV positive no TB HIV negative no TB 1 Characteristics (SD) Men Women Women (n=17) (n=21) Men (n=16) (n=14) Age, years 34.7 (6.5) 29.7 (8.4)b 22.4 (3.2) 24.3 (5.4) Weigh, kg 61.5 (6.3) 57.7 (9.4) 60.5 (6.2) 60.7 (8.7) Height, cm 170.5 (8.2) 155.0 (5.6)a 167.4 (6.6) 159.7 (6.0)b BMI, kg/m2 21.2 (2.2) 24.2 (4.6)b 21.6 (2.3) 23.7 (2.8)b MUAC, mm 27.7 (2.0) 28.5 (3.9) 27.9 (2.1) 29.0 (3.0) Triceps ST, mm 9.8 (7.2) 20.1 (7.2)a 10.5 (4.4) 23.2 (5.7)a Biceps ST, mm 4.5 (2.2) 9.9 (4.4)a 5.2 (2.1) 12.1 (3.9)a Subscapular ST, mm 11.0 (3.8) 19.2 (7.1)a 12.9 (4.2) 17.8 (4.4)a Sacral ST, mm 8.9 (4.2) 15.1 (6.9)b 9.9 (3.9) 16.0 (4.0)a Waist circumference, cm 77.6 (3.5) 77.6 (8.8) 74.6 (5.4) 76.0 (6.1) Hip circumference, cm 91.1 (5.6) 97.0 (9.9)b 90.9 (5.0) 99.3 (6.7)b BIA FFM, kg 51.2 (6.8) 38.2 (3.8)a 49.9 (5.2) 41.3 (6.0)a BIA FFMI, kg/m2 18.3 (1.2) 16.6 (1.2)a 18.6 (1.4) 16.6 (1.1)a WC FFM, kg 52.1 (4.6) 39.4 (5.2)a 52.3 (3.9) 42.7 (5.0)a WC FFMI, kg/m2 18.0 (1.6) 16.5 (2.5) 18.7 (1.5) 16.7 (1.6)b BMI FFM, kg 51.8 (4.6) 39.7 (4.0)a 51.5 (3.8) 43.2 (4.7)a BMI FFMI, kg/m2 17.8 (1.4) 16.6 (1.9)b 18.5 (1.2) 16.9 (1.2)b MUAC FFM, kg 52.5 (5.3) 40.1 (4.4)a 53.1 (3.7) 42.7 (4.6)a MUAC FFMI, kg/m2 18.1 (1.6) 16.7 (2.1) 19.0 (1.6) 16.7 (1.3)a BIA Fat mass, kg 9.1 (3.2) 17.8 (8.2)b 7.9 (2.5) 18.2 (5.4)a BIA FMI, kg/m2 3.2 (1.2) 7.6 (3.7)a 2.9 (1.0) 7.1 (2.1)a WC Fat mass, kg 10.0 (2.4) 18.3 (5.2)a 7.5 (2.2) 18.0 (4.1)a
WC FMI, kg/m2 3.5 (0.8) 7.7 (2.4)a 2.7 (0.8) 7.0 (1.4)a
67
BMI Fat mass, kg 10.3 (2.7) 18.0 (6.3)a 8.2 (2.5) 17.5 (4.3)a BMI FMI, kg/m2 3.6 (1.0) 7.6 (2.9)a 3.0 (1.1) 6.8 (1.6) a MUAC Fat mass, kg 9.6 (2.9) 17.6 (5.7)a 6.6 (2.6) 18.0 (4.6)a MUAC FMI, kg/m2 3.3 (1.1) 7.4 (2.7)a 2.4 (0.9) 7.0 (1.7)a ∑4 SF FFM, kg 28.2 (3.5) 28.3 (6.1) 27.0 (3.1) 30.4 (5.4)b ∑4 SF FFMI, kg/m2 9.7 (1.3) 11.9 (2.9)b 9.7 (1.3) 11.9 (1.8)a ∑4 SF Fat mass, kg 33.9 (3.4) 29.4 (3.3)a 32.7 (2.6) 30.3 (3.4) ∑4 SF FMI, kg/m2 11.7 (1.1) 12.3 (1.8) 11.7 (1.0) 11.9 (1.0)
ªp-value < 0.001, bp-value < 0.05; p-values were obtained with Mann-Whitney test and were comparisons between men and women. 1Values are means ± standard deviation (SD). TB = Tuberculosis disease, WC = waist circumference, MUAC = mid-upper arm circumference, SF = Skinfold thickness, BMI, body mass index, FFM = Fat-free mass, FFMI = Fat-free mass index, and FMI = Fat mass index.
68
Table 4:3 Comparison fat-free mass and fat mass as measured by BIA and by
equations that involved anthropometry measurements for all participants (n=131)
Slope, *p-value 95% limits of Characteristic Correlation bias (se) dBias (SD) agreement
Waist FFm, kg 0.90a -0.04 (0.04) -0.07 (3.77) -7.52, 7.38 0.270
Fat mass, kg 0.83a 0.21 (0.04)a -1.48 (3.31) -7.97, 5.01 0.002 4 Skinfolds FFM, kg 0.43a 0.50 (0.10)a 17.86 (7.38) 3.40, 32.32 <0.001 Fat mass, kg 0.36a 0.78 (0.11)a -19.41 (6.76) -32.66, -6.16 <0.001 BMI FFM, kg 0.93a 0.06 (0.03) -0.50 (2.87) -6.13, 5.13 0.356
Fat mass, kg 0.91a 0.18 (0.03)a -1.05 (2.32) -5.60, 3.50 <0.001 MUAC FFM, kg 0.91a -0.01 (0.04) -1.40 (3.39) -8.04, 5.24 <0.001 Fat mass, kg 0.87a 0.09 (0.04)b -0.15 (2.98) -5.99, 5.69 0.002 ªp-value < 0.001, bp-value < 0.05. Correlations are spearman correlation coefficients. dBias calculated as the mean difference in the BIA and anthropometric measures. 95% limits of agreement calculated as mean difference ± 1.96 x S.D. *P-value for difference obtained by paired t-test. SD = standard deviation, MUAC = mid-upper arm circumference, BMI = body mass index, and FFM = Fat-free mass.
.
69
Table 4:4 Comparison fat-free mass and fat mass measured by BIA and by equation with waist circumference
Slope, 95% limits *p- of value Characteristic Correlation bias (se) dBias (SD) agreement
All men FFM, kg 0.81a 0.17 (0.09) -0.50 (3.53) -7.42, 6.42 0.270 Fat mass, kg 0.82a 0.20 (0.06)a -0.69 (1.66) -3.94, 2.56 0.002 HIV+TB+ men FFM, kg 0.81a 0.31 (0.14) 1.26 (2.57) -3.78, 6.30 0.156 Fat mass, kg 0.68b -0.05 (0.22) -1.60 (1.15) -3.85, 0.65 0.002 HIV-TB+ men FFM, kg 0.88a 0.08 (0.11) 0.58 (2.35) -4.03, 5.19 0.307 Fat mass, kg 0.81a 0.28 (0.16) -0.93 (1.50) -3.87, 2.01 0.017 HIV+TB- men FFM, kg 0.78a 0.43 (0.20)b -0.88 (4.68) -10.05, 8.29 0.448 Fat mass, kg 0.77a 0.34 (0.11)a -0.97 (1.97) -4.83, 2.89 0.059 HIV-TB- men FFM, kg 0.87a 0.29 (0.16) -2.42 (2.96) -8.22, 3.38 0.005 Fat mass, kg 0.87a 0.18 (0.10) 0.44 (1.23) -1.97, 2.85 0.173 All women FFM, kg 0.75a -0.13 (0.09) 0.31 (3.94) -7.41, 8.03 0.518 Fat mass, kg 0.89a 0.54 (0.06)a -2.17 (4.14) -10.28, 5.94 <0.001 HIV+TB+ women FFM, kg 0.79a -0.05 (0.17) 2.45 (2.71) -2.86, 7.76 0.001 Fat mass, kg 0.81a 0.43 (0.11)a -4.97 (2.62) -10.11, 0.17 <0.001
HIV- TB+ women
70
FFM, kg 0.54b 0.35 (0.31) 0.98 (5.41) -9.61, 11.59 0.508 Fat mass, kg 0.66b 0.78 (0.20)a -2.85 (4.91) -12.47, 6.77 0.049
HIV+TB- Women FFM, kg 0.73a -0.35 (0.15)b -1.18 (3.20) -7.45, 5.09 0.108 Fat mass, kg 0.95a 0.49 (0.09)a -0.47 (3.84) -8.00, 7.06 0.579
HIV- TB- women FFM, kg 0.71b 0.19 (0.18) -1.36 (3.41) -8.04, 5.32 0.159 Fat mass, kg 0.72b 0.37 (0.20) 0.19 (3.17) -6.00, 6.38 0.826
ªp-value < 0.001, bp-value < 0.05. Correlations are spearman correlation coefficients. dBias calculated as the mean difference in the BIA and anthropometric measures. 95% limits of agreement calculated as mean difference ± 1.96 x S.D. *P-value for difference obtained by paired t-test. SD = standard deviation, TB- = no TB disease, TB = Tuberculosis disease, HIV- = HIV negative, HIV+ = HIV positive, FFM = Fat-free mass, and FM = Fat mass. .
71
Table 4:5 Comparison fat-free mass and fat mass measured by BIA and by equation with sum of 4 skinfolds
Slope, 95% limits *p- of value Characteristic Correlation bias (se) dBias (SD) agreement
All men FFM, kg 0.69a 0.42 (0.11)a 24.03 (4.4) 15.5, 32.6 <0.001 Fat mass, kg 0.89a -0.03 (0.10) -25.23 (2.23) -29.6, -20.9 <0.001 HIV+TB+ men FFM, kg 0.67b 0.57 (0.17)a 25.45 (3.39) 18.8, 32.1 <0.001 Fat mass, kg 0.70b -0.42 (0.19) -25.79 (1.48) -28.7, -22.9 <0.001 HIV-TB+ men FFM, kg 0.85a 0.44 (0.12)a 25.32 (2.86) 19.7, 30.9 <0.001 Fat mass, kg 0.81a -0.28 (0.15) -25.67 (1.65) -28.9, -22.4 <0.001 HIV+TB- men FFM, kg 0.52b 0.83 (0.27)a 22.97 (5.94) 11.3, 34.6 <0.001 Fat mass, kg 0.50b -0.07 (0.28) -24.83 (3.14) -31.0, -18.7 <0.001 HIV-TB- men FFM, kg 0.64b 0.59 (0.24)b 22.82 (4.05) 14.9, 30.8 <0.001 Fat mass, kg 0.71b -0.02 (0.23) -24.81 (2.02) -28.8, -20.9 <0.001 All women FFM, kg 0.68a -0.27 (0.10)a 12.48 (4.79) 3.1, 21.9 <0.001 Fat mass, kg 0.86a 0.79 (0.06)a -14.34 (5.08) -24.3, -4.4 <0.001 HIV+TB+ women FFM, kg 0.76a -0.25 (0.18) 14.88 (3.26) 8.5, 21.3 <0.001 Fat mass, kg 0.71a 0.70 (0.12)a -17.40 (3.33) -23.9, -10.9 <0.001
HIV-TB+ women
72
FFM, kg 0.44 0.33 (0.36) 14.37 (6.08) 2.5, 26.3 <0.001 Fat mass, kg 0.59b 0.93 (0.26)a -16.23 (5.61) -27.2, -5.2 <0.001 HIV+TB- Women FFM, kg 0.52b -0.56 (0.19)a 9.88 (4.53) 1.0, 18.8 <0.001 Fat mass, kg 0.92a 0.86 (0.09)a -11.53 (5.20) -21.7, -1.3 <0.001 HIV-TB- women FFM, kg 0.77b 0.12 (0.16) 10.91 (3.6) 4.7, 17.1 <0.001 Fat mass, kg 0.77b 0.48 (0.16)a -12.08 (3.08) -18.1, -6.0 <0.001 ªp-value < 0.001, bp-value < 0.05. Correlations are spearman correlation coefficients. dBias calculated as the mean difference in the BIA and anthropometric measures. 95% limits of agreement calculated as mean difference ± 1.96 x S.D. *P-value for difference obtained by paired t-test. SD = standard deviation, TB- = no TB disease, TB = Tuberculosis disease, HIV- = HIV negative, HIV+ = HIV positive, FFM = Fat-free mass, and FM = Fat mass.
73
Table 4:6 Comparison fat-free mass and fat mass measured by BIA and by equation with BMI
Slope, 95% limits *p- of value Characteristic Correlation bias (se) dBias (SD) agreement
All men FFM, kg 0.87a 0.20 (0.07)b -0.35 (2.90) -6.03, 5.33 0.356 Fat mass, kg 0.86a 0.04 (0.05) -0.85 (1.34) -3.48, 1.78 <0.001 HIV+TB+ men FFM, kg 0.88a 0.20 (0.10) 0.77 (1.84) -2.84, 4.38 0.218 Fat mass, kg 0.85b 0.05 (0.16) -1.11 (0.08) -2.67, 0.46 0.002 HIV-TB+ men FFM, kg 0.89a 0.11 (0.11) 0.42 (2.22) -3.93, 4.77 0.429 Fat mass, kg 0.75a 0.00 (0.18) -0.77 (1.68) -4.06, 2.52 0.068
HIV+TB- men FFM, kg 0.85a 0.41 (0.15)b -0.58 (3.89) -8.20, 7.04 0.547 Fat mass, kg 0.92a 0.17 (0.07)b -1.28 (1.27) -3.77, 1.21 0.001 HIV-TB- men FFM, kg 0.82a 0.31 (0.13)b -1.66 (2.53) -6.62, 3.30 0.019 Fat mass, kg 0.81a 0.01 (0.10) -0.33 (1.16) -2.60, 1.94 0.279 All women FFM, kg 0.83a 0.01 (0.07) -0.63 (2.85) -6.22, 4.96 0.067 Fat mass, kg 0.96a 0.37 (0.03)a -1.22 (2.92) -6.94, 4.50 0.001 HIV+TB+ women FFM, kg 0.76a -0.08 (0.15) 0.44 (2.49) -4.44, 5.32 0.432 Fat mass, kg 0.90a 0.38 (0.09)a -2.96 (2.37) -7.61, 1.69 <0.001
74
HIV-TB+
women FFM, kg 0.68b 0.30 (0.19) 0.30 (3.73) -7.01, 7.61 0.766
Fat mass, kg 0.96b 0.50 (0.09)a -2.16 (3.15) -8.33, 4.01 0.023 HIV+TB-
Women FFM, kg 0.87a -0.06 (0.11) -1.51 (1.87) -5.18, 2.16 0.001 Fat mass, kg 0.98a 0.28 (0.05)a -0.14 (2.36) -4.77, 4.49 0.789 HIV-TB- women FFM, kg 0.81a 0.26 (0.15) -1.87 (2.96) -7.67, 3.93 0.035 Fat mass, kg 0.78b 0.30 (0.15) 0.69 (2.55) -4.31, 5.69 0.329
ªp-value < 0.001, bp-value < 0.05. Correlations are spearman correlation coefficients. dBias calculated as the mean difference in the BIA and anthropometric measures. 95% limits of agreement calculated as mean difference ± 1.96 x S.D. *P-value for difference obtained by paired t-test. SD = standard deviation, TB- = no TB disease, TB = Tuberculosis disease, HIV- = HIV negative, HIV+ = HIV positive, FFM = Fat-free mass, and FM = Fat mass.
75
Table 4:7 Comparison fat-free mass and fat mass measured by BIA and by equation with MUAC
Slope, *p- 95% limits of value Characteristic Correlation bias (se) dBias (SD) agreement
All men FFM, kg 0.80a 0.21 (0.08)b -2.03 (3.32) -8.54, 4.48 <0.001 Fat mass, kg 0.82a -0.08 (0.07) 0.83 (2.01) -3.11, 4.77 0.002 HIV+TB+ men FFM, kg 0.70b 0.36 (0.16) -1.83 (2.86) -7.44, 3.78 0.073
Fat mass, kg 0.67b -0.07 (0.27) 1.49 (1.37) -1.20, 4.18 0.007 HIV-TB+ men FFM, kg 0.82a 0.04 (0.12) -1.74 (2.59) -6.82, 3.34 0.011 Fat mass, kg 0.65b 0.01 (0.22) 1.39 (2.06) -2.65, 5.43 0.011 HIV+TB- men FFM, kg 0.79a 0.26 (0.16) -1.30 (3.97) -9.08, 6.48 0.195 Fat mass, kg 0.83a 0.12 (0.12) -0.55 (1.85) -4.18, 3.08 0.236 HIV-TB- men FFM, kg 0.64b 0.39 (0.20) -3.25 (3.53) -10.17, 3.67 0.002 Fat mass, kg 0.70b 0.11 (0.17) 1.26 (1.89) -2.44, 4.96 0.018 All women FFM, kg 0.77a 0.05 (0.09) -0.85 (3.37) -7.46, 5.76 0.038 Fat mass, kg 0.91a 0.38 (0.05)a -1.01 (3.41) -7.69, 5.67 0.016 HIV+TB+ women FFM, kg 0.70a -0.02 (0.18) 0.17 (2.74) -5.20, 5.54 0.781 Fat mass, kg 0.72a 0.36 (0.12)b -2.69 (2.69) -7.96, 2.58 <0.001
HIV-TB+ women
76
FFM, kg 0.52 0.45 (0.28) -0.35 (5.04) -10.23, 9.53 0.797 Fat mass, kg 0.82b 0.67 (0.18)b -1.51 (4.50) -10.33, 7.31 0.232
HIV+TB- Women FFM, kg 0.80a -0.16 (0.13) -1.87 (2.28) -6.34, 2.60 0.001 Fat mass, kg 0.99a 0.38 (0.05)a 0.22 (2.89) -5.44, 5.88 0.730
HIV-TB- women FFM, kg 0.77b 0.29 (0.17) -1.35 (3.38) -7.97, 5.27 0.159 Fat mass, kg 0.79a 0.22 (0.18) 0.18 (2.95) -5.60, 5.96 0.824
ªp-value < 0.001, bp-value < 0.05. Correlations are spearman correlation coefficients. dBias calculated as the mean difference in the BIA and anthropometric measures. 95% limits of agreement calculated as mean difference ± 1.96 x S.D. *P-value for difference obtained by paired t-test. SD = standard deviation, TB- = no TB disease, TB+ = Tuberculosis disease, HIV- = HIV negative, HIV+ = HIV positive, FFM = Fat-free mass, and FM = Fat mass.
77
CHAPTER 5
INDICATORS OF DIETARY ADEQUACY AMONG HIV POSITIVE AND HIV
NEGATIVE ADULTS WITH OR WITHOUT TUBERCULOSIS IN URBAN
KAMPALA, UGANDA
84
Abstract
Objective To assess nutritional adequacy of dietary intake and validity of simple low-
cost methods to evaluate nutritional adequacy of diets consumed among HIV positive and
HIV negative adults with or without tuberculosis in urban Uganda
Methods In a cross-sectional study of 131 HIV positive and HIV negative adults with or
without tuberculosis, 24-hour dietary intake recall was assessed. Two different dietary
diversity indices were created: food variety score (FVS), a simple count of items, and diet
diversity score (DDS), a count of food groups. Mean adequacy ratio (MAR) of intake to
recommended intake (each truncated at one) of energy and ten nutrients, was calculated
as an indicator of nutrient adequacy.
Results All participants (100%) consumed at least cereals, roots, and tubers, and 90% consumed vegetables not rich in vitamin A such tomatoes and onions while only 45% consumed vitamin-A-rich fruits and vegetables, and only 15% consumed eggs. The mean
FVS and DDS for the study population were low 8.1 ± 2.8 and 4.7 ± 1.4, respectively.
Both men and women regardless of tuberculosis and HIV status, had carbohydrate and ascorbic acid deficiency in the range of 0 to 30% whereas other nutrient intakes including energy, protein, dietary fiber, calcium, magnesium, zinc, iron, vitamin A, vitamin D, and folate had deficiencies ranging 25% to 100%. When a MAR of 65% was used as a cut-off
point for nutrient adequacy, it was found that FVS must be 9 or more and DDS must at
least 5. Among women, both FVS and DDS had a high ability to identify participants
with an inadequate or adequate diet while among men FVS had a high ability to identify
individuals with inadequate diet but low ability to identify those with adequate diet, DDS
85
had low ability to identify individuals with inadequate diet but had a high ability to identify those with adequate diet.
Conclusion The dietary consumption in this study population was monotonous, rich in carbohydrates and deficient in nutrients regardless of gender, tuberculosis, and HIV status. The ability of FVS and DDS indices to identify individuals with inadequate or adequate diet consumption differed by gender. The FVS was a better predictor of nutritional adequacy among women whereas DDS was among men. Validation in follow- up studies and in different regions is needed.
86
Background
Micronutrient malnutrition remains one of the largest nutritional problems worldwide,
affecting people in both developed and developing countries (WHO 2002). Among
individuals with tuberculosis and among individuals with HIV infection, a number of
micronutrients deficiencies have been described (van Lettow, Fawzi, and Semba 2003).
Moreover, micronutrient deficiencies not only impair host immune functions, but may
also affect efficacy of tuberculosis drugs (Thurnham 2004), and impair weight gain
which may be important to survival (Paton et al. 2004). The deficiencies may be largely
due to a habitually low dietary consumption of micronutrients in relation to requirements
or due to anorexia (Hopewell P.C. 1994), impaired absorption of nutrients or increased
catabolism.
Dietary diversification is central and sustainable to food based intervention strategies for
combating multiple micronutrient deficiencies and total energy intake in developing
countries (Tontisirin, Nantel, and Bhattacharjee 2002). Among the foremost steps and
important element in planning dietary diversification strategies is dietary assessment.
However, validity studies for simple low-cost assessment methods to evaluate the
nutritional adequacy of diets across different cultural settings, disease conditions, and age
groups are still lacking in sub-Saharan Africa where there is a high burden of multiple
micronutrient deficiencies and tuberculosis with HIV co-infection. The present cross- sectional study was conducted to fill this gap by assessing adequacy of dietary intake and
87
the relationship between food variety or dietary diversity scores and the nutritional adequacy of the diet among HIV positive and HIV negative adults with or without tuberculosis in urban Kampala, Uganda.
Methods
In a cross-sectional study, 132 participants with age 18 or more years residing in
Kampala district or 20 km from the study site if residence was outside Kampala in
Uganda were enrolled. Data collection was conducted between November 2007 and
March 2008, a period that coincides with harvesting and light rains in November and
December and dry season in January and February. One participant was excluded from the analysis because of prior tuberculosis treatment. The study was conducted at the
National Tuberculosis and Leprosy Program (NTLP) Clinic of the national tertiary teaching hospital, Mulago complex. Of the 131 participants who were included in the analysis, 63 were tuberculosis patients who were recruited at the Mulago NTLP Clinic;
38 were HIV positive patients without TB and recruited at the Infectious Disease Institute
Clinic (IDI) located 500 meters from the Mulago NTLP Clinic; and 30 were HIV negative individuals without tuberculosis from the community where enrolled tuberculosis patients resided. The institutional review boards at Case Western Reserve
University and Joint Clinical Research Center approved the study, with final approval by the Uganda National Council for Science and Technology. All participants provided written informed consent to the study.
88
All subjects in the study were given appropriate pre- and post-test HIV counseling and
AIDS education. HIV-1 infection was diagnosed on the basis of a positive enzyme-linked immunosorbent assay for HIV-1 antibodies (Recombigen; Cambridge Biotech,
Cambridge, MA). At enrollment, basic demographic information and a medical history were collected, and a standardized physical examination was conducted by a medical officer. Active pulmonary TB was confirmed by sputum smear microscopy and culture.
Patients with active TB were treated with standard four-drug chemotherapy for tuberculosis per guidelines of the Ugandan Ministry of Health. Similarly, HIV positive patients eligible for antiretroviral therapy were started on treatment and cotrimoxazole prophylaxis at the IDI clinic. Anthropometric measurements included height and weight.
Body-mass index (BMI) was computed using the relationship of weight in kilograms
divided by height in meters squared (kg/m2). Weight was taken using Hanson digital
electronic bathroom weighing scales to the nearest 100g. Height was measured to the
nearest mm by calibrated standing height using a stadiometer. All anthropometric
measurement values were the mean of duplicates.
The dietary intake assessment was made using a single 24-hour dietary recall with open
ended questions. The reference period for the 24-hour recall was the day prior to the day of the interview. In all instances, the interview was held only with the interviewee, no one
else was present except for children. The questionnaire was pre-tested by administering it
to 8 individuals selected randomly from the neighboring community to the study site. The
assessment was conducted by four trained study nursing staffs and continuously
supervised by a nutritionist using local food photographs, portion-size images, and
89
volumetric vessels to increase the accuracy of the recall. The nutritive value of raw ingredients was computed using the East African food composition table database whose database was imported into the NutriSurvey software (http:www.nutrisurvey.de) to ease the computations. The database used was predominantly for local Ugandan diet. When the East African food composition table was found deficient in certain food items, the
United States Department of Agriculture database and the African composition table were used.
We evaluated the dietary adequacy in this study population basing methods that have been described previously (Hatloy, Torheim, and Oshaug 1998). The dietary diversity score (DDS) was defined as the number of food groups consumed over a period of 24 hours. The diet was classified according to nine food groups as recommended by Food and Agricultural Organization (FAO) (Table 5:3). Other remaining food items such as tea, sugar, salt, sweets, spices, commercial energy drinks, and alcohol were not used in the DDS and food variety score (FVS) calculations. The FVS was defined as the number of food items consumed over a 24-hour period, from a possible total of 127 Items. The possible total (n=127) reflects all the difference types of food items eaten by this sample population in a 24-hour period.
To estimate the nutrient adequacy of the diet, we calculated the nutrient adequacy ratio
(NAR %) for 10 micronutrients, energy, protein, fat, carbohydrate, and dietary fiber. The
NAR for a given nutrient is the ratio of a participant’s intake to the daily recommended
90
allowance for the participant’s sex. The Food and Agriculture Organization/World Health
Organization 2002 Human Vitamin and Mineral requirements (FAO/WHO 2002) were
used for vitamin A, vitamin B6, vitamin C, thiamin, riboflavin, folate, magnesium,
calcium, iron, and zinc whereas energy, fat, protein, carbohydrate, and dietary fiber; the
Panel on Macronutrients, Panel on the Definition of Dietary Fiber, Subcommittee on
Upper Reference Levels of Nutrients, Subcommittee on Interpretation and Uses of
Dietary Reference Intakes Food and Nutrition Board (Dietary Reference Intake for
Energy, Carbohydrate, Fiber, Fat, Fatty Acids, Cholesterol, Protein, and Amino Acids
(Macronutrients). A report of the Panel on Macronutrients, Subcommittees on Upper
Reference Levels of Nutrients and Interpretation and Uses of Dietary Reference Intakes, and the Standing Committee on the Scientific Evaluation of Dietary Reference Intakes
2005) was used (Table 5:4). The recommendation for adequate intake was used for vitamin D (Atkinson and Ward 2001). In the case of iron and zinc, the category for
moderate bioavailability was used. The mean adequacy ratio (MAR %) was calculated as
a measure of the adequacy of the overall diet, where MAR is the sum of each NAR
(truncated at 100%) divided by the number of nutrients (excluding energy, fat, protein,
carbohydrates, and dietary fiber) (Madden, Goodman, and Guthrie 1976; Hatloy,
Torheim, and Oshaug 1998).
MAR (Mean Adequacy Ratio) = ∑NAR (each truncated 100%)
Number of nutrients
91
NAR was truncated at 100% so that a nutrient with a high NAR could not compensate for
a nutrient with a low NAR. For both NAR and MAR a value of 100% is the ideal since it
means that the intake is the same as the requirement.
Analytic strategy
All study participants in the analysis were categorized into 4 mutually independent
groups: HIV positive patients with and without tuberculosis disease, HIV negative
patients with and without tuberculosis disease. Spearman’s rank correlation tests were
performed between NARs or MAR and FVS or DDS. Measures of central tendency and
variability were compared between men and women across 4 mutually exclusive groups
using using Wilcoxon-Mann-Whitney test for average age, weight, height, BMI, FVS,
DDS, and MAR. A non-parametric test was used because of small size issues in
subgroups and nearly all parameters were not normally distributed even after log
transformations. We evaluated FVS and DDS for sensitivity and specificity with MAR as
the ‘gold standard’ of nutritionally adequate intake. Sensitivity is the proportion of
positives that are correctly identified by the test, while specificity is the proportion of
negatives that are correctly identified by the test (Altman D.G 1997). Positive is defined
as FVS or DDS below a given cut-off point. Different cut-off points were tested to find the levels of FVS and DDS that would give high sensitivity without losing too much specificity. Those with a nutritionally inadequate diet, defined as MAR below a certain cut-off point, and FVS or DDS below the cut-off point were defined as true positives.
Those with a nutritionally adequate diet, or MAR greater than the cut-off point, and FVS
92
or DDS above the cut-off were defined as true negatives. We used linear regression to estimate MAR scores for different levels of FVS and DDS. All analyses were performed using SAS version 9.2 Cary software, (North Carolina SAS Institute Inc 2004) whereas the receiver operator curves plotted in Microsoft Excel 2007.
Results
Of the 131 participants analyzed, 31 were HIV positive with tuberculosis, 32 were HIV negative with tuberculosis, 38 were HIV positive without tuberculosis, and 30 were HIV negative without tuberculosis. Overall men and women in the study population had similar age except among HIV positive individuals without tuberculosis where men were significantly older than women. The average height was significantly higher among men compared to women regardless of tuberculosis and HIV status (Table 5:1 and 5:2).
A total of 127 different food items were eaten by all adults (n=131) in the study population, corresponding to a theoretical maximum of 127 FVS. The mean FVS was 8.1
± 2.8 with a minimum score of 1.0 and maximum of 15.0. The mean DDS value for the total sample was 4.7 ± 1.4 with a minimum of 1.0 and maximum of 8.0. There were no significant differences in average FVS, DDS, and MAR between men and women regardless of tuberculosis and HIV status except for FVS among HIV negative individuals without tuberculosis; women had significantly lower average FVS compared to men (Table 5:1 and 5:2). Of note, women had lower magnitudes of FVS, DDS, and
MAR compared to men regardless of tuberculosis and HIV status.
93
All participants in the study sample had consumed some kind of cereal, roots, or tubers,
particularly local green plantain, maize meal, sweet potatoes, or rice (Table 5:3). Food
items that were consumed by more than 80% of the sample were other vegetables not rich
in vitamin A including tomatoes, onions, green pepper, and eggplant. Only 32% of all
participants consumed vitamin A rich fruits and vegetables such as green vegetables,
carrot, mangoes, and papaya (locally called pawpaws). The least consumed food groups
included eggs (15%), dairy/dairy products (25%), and other fruits that are not rich in
vitamin A (45%) (Table 5:3). More than 60% of the sample consumed at least legumes
and nuts (62%); meat, poultry, and fish (67%); and fats and oils (65%).
The proportion of participants with a nutrient intake below the recommended daily
allowance varied between nutrients (Table 5:4). There were no nutrients, for which
participants had sufficient nutrient intake, i.e., for which the nutrient adequate ratio
(NAR) was 100% for all the participants. Nutrients such as carbohydrate and vitamin C
had median NAR above 100%; however, a substantial proportion of participants had
intake below the recommended allowance. Nutrients for which nutrient intake for
participants were 80% or more below the recommended daily allowance included vitamin
D, calcium, iron, and percent protein of energy (Table 5:4).
There were no significant differences in 24-hour dietary nutrient intake deficiencies between men and women for most nutrients except total fat, iron, and folate among HIV
94
negative individuals without TB; women had a greater proportion of deficiencies
compared to men (Tables 5:5 and 5:6). Women had a deficiency of 57% versus 13%,
100% versus 69%, and 71% versus 25% for total fat, iron, and folate, respectively
compared to men.
All nutrient adequacy expressed as NAR for the different nutrients correlated positively
with Food Variety Score (FVS) and Dietary Density Score (DDS) among both women
and men except percent carbohydrate NAR. For example, the correlation between energy
intake NAR and FVS was 0.69 and it was 0.56 with DDS among women (Table 5:7). All
correlations were significant except percent protein of energy NAR with FVS and DDS
regardless of gender, percent fat of energy NAR with FVS among women and men,
percent carbohydrate of energy with FVS and DDS among women, calcium and vitamin
D NARs with FVS among men. There was also strong positive correlation between FVS
or DDS and the Mean Adequacy Ratio (MAR), the overall score for nutritional adequacy
of the diet (Table 5:7).
The mean MAR value was 63 ± 23 among women and 71 ± 18 among men (Table 5:7).
The ideal cut-off for nutrient adequacy should be 100. This would mean that all nutrients were consumed in adequate amounts. In the present study, none of the participants attained this level. Among women, 59% had MAR greater than or equal to 60; 47% of the women participants had MAR greater than or equal to 65; and 39% had MAR greater
than or equal to 75. Whereas among men, 74% had MAR greater than or equal to 60;
95
59% of the men participants had MAR greater than or equal to 65; and 44% had MAR
greater than or equal to 75.
We tested different cut-off points of FVS for sensitivity and specificity against different
definitions of a nutritionally adequate diet ranging from MAR = 60 to MAR = 75 as
shown in Figures 5:1 and 5:2 receiver operator curves. We performed this in order to find
an optimal cut-off point for FVS that can identify the maximum inadequate diets as inadequate (high sensitivity), without losing too much ability to identify those with a nutritionally adequate diet (specificity). The figures show that among women and men the sensitivity and specificity were not much influenced by changing the cut-off points of
MAR. We therefore used MAR = 65 as cut-off point for a nutritionally adequate diet. If we were to get a sensitivity of FVS higher than or equal to 75% among women, the cut- off for FVS must be 9 or higher, while to get a specificity of at least 25%, the cut-off for
FVS needs to be 11 or lower. Whereas among men, the cut-off for FVS must be 9 or
higher to get a sensitivity higher than or equal to 75%, while to get a specificity of at least
25%, the cut-off for FVS needs also to be 11 or lower. With a cut-off for FVS on 9 among women and on 9 among men, the sensitivity will be 76% for women and 76% for men and the specificity 70% for women and 67% for men, respectively. If the cut-off for
FVS had been increased by one, the sensitivity would have been increased to 92% among women and compared to 80% among men, but the specificity would decrease to 45% among women and to 50% among men. On the other hand, if the cut-off had decreased by one, the sensitivity would have dropped to 73% among women and 48% among men, and the specificity would have increased to 88% among women and to 83% among men.
96
We undertook the same exercise to decide the cut-off point for DDS as shown in Figures
5:3 and 5:4. A cut-off point for DDS among women of 5 gave a sensitivity of 70% and a
specificity of 85%. Whereas among men, a cut-off point of 5 gave a sensitivity of 64% and a specificity of 83%. If the cut-off point had been increased to 6 among women, the sensitivity would have been 95% and the specificity of 48% whereas among men the sensitivity would have been 96% and specificity 39%. Decreasing the cut-off point to 4 among women, the sensitivity would have decreased to 32% and the specificity increased to 97% whereas among men the sensitivity would have decreased to 16% and specificity increased to 100%. We therefore took a cut-off point for DDS to be 5 in the present study as the only one point that gives both sensitivity higher than 64% and specificity higher than 80% among both women and men.
Our data shows an increasing MAR with increasing DDS and FVS (Table 5:8). The FVS needs to be 7 to 9 to give a MAR above 65 whereas DDS needs to be at least 5 and above to give satisfactory MAR for women and men. The theoretical estimates of linear regression appear to confirm the data (Table 5:9). To achieve a MAR greater than or equal to 65 with a DDS of 5, one will need FVS of at least 9 among women and 5 among men. For a DDS of 7, the FVS would be 7 among women and as few as 3 among men.
The results from the regression model (Table 5:9, foot note) show that it is only FVS that contributes significantly to the fit of the model among women whereas DDS did fit among men even after adjusting for age, HIV status,
97
Discussion
The present cross-sectional study that analyzed 131 HIV positive and HIV negative
adults with or without tuberculosis was conducted to assess nutritional adequacy of
dietary intake and validity of simple low-cost methods to evaluate nutritional adequacy of diets consumed among adults in urban Uganda. The dietary consumption in this study population was of low variety and diversity regardless of gender; composed of mostly cereals, roots, tubers, and vegetables not rich in vitamin A. The dietary consumption was
deficient in most nutrients except carbohydrates, vitamin C and B6 that had more than
75% of the study population with 1 nutrient adequate ratio. The deficiency was not
affected by gender, tuberculosis, and HIV status except for iron and folate that differed
by gender among HIV negative individuals. The FVS and DDS increase were associated
with positive increase for nearly all nutrient intake adequacy ratios. The ability of FVS
and DDS indices to identify individuals with inadequate or adequate diet consumption
differed by gender. Prediction of nutritional adequacy differed by gender. The simple
counting of food items (FVS) was a better predictor of nutritional adequacy of the diet
among women whereas counting food groups (DDS) was among men. Thus, FVS and
DDS can provide a rapid and efficient means to health workers to quickly estimate
nutritional adequacy of the diet and also to monitor consumption during management of
tuberculosis and HIV.
Although the interpretation of findings in the present study may be limited by the cross-
sectional nature of the design, specifically use of a single 24-hour dietary recall that the
98
nutrition intakes do not reflect health endpoints (Vucic et al. 2009) and overall dietary habits (Biro et al. 2002), we employed local utensils with known size and food photographs to enhance recall and estimation of amount for the intake. Further, the dietary recall overlapped with one of the weekend.
Findings in the present study suggest that dietary consumption in this study population was monotonous, rich in carbohydrates and deficient in nutrients. All participants (100%) consumed at least cereals, roots, and tubers, and 90% consumed vegetables not rich in vitamin A such tomatoes and onions while only 45% consumed vitamin-A-rich fruits and vegetables, and only 15% consumed eggs. The mean FVS and DDS for the study population were only 8.1 ± 2.8 and 4.7 ± 1.4, respectively. Both men and women regardless of tuberculosis and HIV status, had carbohydrate and ascorbic acid deficiency in the range of 0 to 30% whereas other nutrient intakes including energy, protein, dietary fiber, calcium, magnesium, zinc, iron, vitamin A, vitamin D, and folate had deficiencies ranging 25% to 100%. Of note, vitamin D had the worst deficiency range from 90% to
100%. This substantial deficiency in vitamin D may a risk to reactivation of tuberculosis
(Sita-Lumbsden A et al. 2007). It is probably difficult to compare results in the present study with other countries on the basis of FVS and DDS because of the different ways in which these indicators have been defined and calculated in different countries (Hatloy,
Torheim, and Oshaug 1998). Further, studies are still limited in populations from sub-
Saharan Africa. Nevertheless, our findings corroborate with the cross-sectional study among the elderly that was conducted in South Africa (Oldewage-Theron and Kruger
2008). Findings in this study revealed low FVS and DDS, and a dietary consumption that
99
was rich in carbohydrate and nutrient. The strength of our study; however, was the presence of full panel of control groups that comprised of HIV positive and HIV negative adults with or without tuberculosis.
The monotonous diet in this study population reflects on the subsistence economy in
Uganda. Although this study was conducted in urban setting, it appears people purchase mostly agricultural food stuffs similar to the staple food that is grown in their rural home steady. Further, due to lack of public health nutrition education in the general population, people do not know the vital importance of food variety and diversity. People who live at subsistence levels often have no choice but to consume monotonous diets that are poor in nutrients, resulting in poor diet quality (Cannon G 2001). Further, the possible poverty level in this population expose to the monotonous diet. Poor populations suffer most in achieving dietary diversity because of they consume a usual diet based on starchy staple foods containing little or no animal or diary products and few fruit and vegetables, resulting in multiple nutrient deficiencies (Ruel 2003).
Our results show that the FVS and DDS can identify fairly adults with an inadequate nutrient intake. With cut-off points for FVS at 9 and DDS at 5 among women, the indices have high ability to identify those with a nutritionally inadequate and adequate diet whereas among men the indices have lower ability to identify those with inadequate diet, but high ability to identify those with adequate diet. To our knowledge, this is the first study to show gender differences in the ability of FVS and DDS indices to identify
100
individuals or populations with adequate and inadequate diet. Although the value of FVS
and DDS have been established in children (Hatloy, Torheim, and Oshaug 1998; Steyn et
al. 2006; Ruel 2003), limited research has been done in other population groups and disease conditions to explore the consequences of low dietary variety on nutritional status and response during treatment, for example in tuberculosis. This study attempted to define dietary variety across a heterogeneous population with or without tuberculosis and
HIV.
For nutrition promotion and prevention of malnutrition in the population of HIV positive and HIV negative with or without tuberculosis, a high sensitivity is desirable to identify accurately most subjects at nutritional risk (Habicht, Meyers, and Brownie 1982), as long as false positives do not cause other risks. False positives do not pose any risk in the case of the present because of the high prevalence of nutrient deficiencies. In assessment of how select optimal cut-off points for FVS and DDS for an adequate diet, the choice of cut-off point for MAR did not have a strong influence. In our analysis, changing the cut-
off point for MAR from 60 to 75 did not have any vital importance for the conclusions.
We used the liberal 65% as the cut-off point for MAR as used in prior studies (Oldewage-
Theron and Kruger 2008; Steyn et al. 2006). This is supported by the results of Tables 6
and 7, where it is shown that to reach a MAR of 65, one needs FVS and DDS of at least 9
and 5 or higher, respectively.
101
The gender differences in the ability of FVS and DDS indices to identify individuals or populations with adequate and inadequate diet could be explained by the cultural factors that may compromise intake among women. For example, unequal distribution of food within households (Carloni 1981; de Hartog A.P 1972), or men may have the opportunity to eat a wider variety or better quality foods outside the home, such as cafes or local restaurants (Holmboe-Ottesen G and Wandel M 1991). The unequal distribution of food within households result from several factors such as women may be trained to show restraint in eating, to give the best foods to men, or to allow others in the family to eat first (Lado 1992; O'Laughlin B 1974; Rosenberg E.M 1980; Dey J 1981).
The present study revealed that diet consumption among HIV positive and HIV negative adults with or without tuberculosis was monotonous, rich in carbohydrates and deficient in nutrients. The ability of FVS and DDS indices to identify individuals with inadequate or adequate diet consumption differs by gender. The simple counting of food items (FVS) was a better predictor of nutritional adequacy of the diet among women whereas counting food groups (DDS) was among men. Thus, FVS and DDS can provide a rapid and efficient means to health workers to quickly estimate nutritional adequacy of the diet, to monitor dietary intake in management of tuberculosis and HIV, and to assess patient’s nutritional knowledge and thereafter provide health education. Validation studies are; however, needed in follow-up studies and other regions of the country.
102
Acknowledgements
We thank all study staff members of the Case Western Reserve University and Makerere
University research collaboration at the Tuberculosis Research Unit in the United States and in Uganda for their assistance; the faculty of staff at Case Western Reserve
University Department of Epidemiology and Biostatistics for the guidance in analyzing the project; and the Fogarty International Center, for the continued support.
This study was supported in part by the AIDS International Training Research Program,
Fogarty International Center, Grant No. TW000011, based at Case Western Reserve
University, Department of Epidemiology and Biostatistics. This work was part of Ezekiel
Mupere’s PhD thesis at Case Western Reserve University.
103
Table 5:1 Characteristics of adult participants with tuberculosis in Kampala,
Uganda
Characteristic HIV positive with TB HIV negative with TB
[mean, (SD)] (n=31) (n=32)
Women Men Women Men
(n=21) (n=10) (n=14) (18)
Age in yrs 29.2 (5.9) 30.9 (4.6) 26.0 (5.4) 26.0 (7.3)
Weight in kg 46.8 (7.3) 54.3 (6.1)b 50.0 (7.8) 53.2 (6.5)
Height in cm 158.9 (6.7) 171.8 (9.3)a 157.4 (13.2) 171.1 (6.0)a
BMI in kg/m2 18.6 (2.9) 18.4 (1.7) 20.5 (4.4) 18.2 (2.0)
FVS 7.1 (2.8) 6.4 (1.7) 8.7 (2.8) 8.7 (2.7)
DDS 4.5 (1.7) 4.7 (1.5) 5.4 (1.6) 4.7 (1.1)
MAR 61.4 (24.3) 68.0 (17.6) 68.7 (17.3) 70.6 (17.4)
ap-value <0.004, bp-value <0.05. FVS = Food Variety Score: the number of all food items used in the study period (1 – 127). DDS = Dietary Diversity Score: number of food groups eaten (1 – 8): cereals, roots, and tubers; vitamin A-rich fruits and vegetables; other fruits not rich in vitamin A; other vegetables not rich in vitamin A; legumes and nuts; meat, poultry, and fish; fats and oils; diary and products; and eggs. MAR = Mean Adequacy Ratio: ratio of 11 micronutrients: vitamin A, vitamin B6, vitamin C, vitamin D, thiamin, riboflavin, folate, magnesium, calcium, iron, and zinc.
104
Table 5:2 Characteristics of adult participants without tuberculosis in Kampala,
Uganda
Characteristic HIV positive with no TB HIV negative with no TB
[mean, (SD)] (n=38) (n=30)
Women Men Women Men
(n=21) (17) (n=14) (16)
Age in yrs 29.7 (8.4) 34.7 (6.5)b 24.3 (5.4) 22.4 (3.2)
Weight in kg 57.7 (9.4) 62.1 (6.6) 60.7 (8.7) 59.8 (5.6)
Height in cm 155.0 (5.6) 170.5 (7.9)a 159.7 (6.0) 167.2 (6.8)b
BMI in kg/m2 24.2 (4.6) 21.4 (2.3)b 23.7 (2.8) 21.4 (2.3)b
FVS 8.2 (3.4) 8.5 (2.4) 7.6 (2.5) 9.6 (2.6)
DDS 4.3 (1.4) 4.8 (1.0) 4.5 (1.5) 5.1 (1.1)b
MAR 60.2 (27.3) 68.5 (17.1) 62.6 (19.5) 74.3 (20.0)
ap-value <0.004, bp-value <0.05. FVS = Food Variety Score: the number of all food items used in the study period (1 – 127). DDS = Dietary Diversity Score: number of food groups eaten (1 – 8): cereals, roots, and tubers; vitamin A-rich fruits and vegetables; other fruits not rich in vitamin A; other vegetables not rich in vitamin A; legumes and nuts; meat, poultry, and fish; fats and oils; diary and products; and eggs. MAR = Mean Adequacy Ratio: ratio of 11 micronutrients: vitamin A, vitamin B6, vitamin C, vitamin D, thiamin, riboflavin, folate, magnesium, calcium, iron, and zinc.
105
Table 5:3 Food groups and food items from 24-hour dietary intake recall among
HIV positive and HIV negative adults in Kampala, Uganda (n=131)
Food groups Frequency (%) % Food item
Cereals, roots, and 100.0 31.7 Green plantain (Musa acuminata) tubers 10.3 Maize meal (Zea mays)
9.9 Sweet potatoes (Ipomoea batatas)
9.0 Rice (Oryza sativa)
Vitamin-A-rich fruits 31.6 34.2 Green vegetables (Amaranthus) and vegetables 23.7 Carrot (Daucus carota)
26.3 Mango fruit (Mangifera)
13.2 Pawpaw (Papaya, Carica)
Other fruits not rich in 44.5 41.7 Passion fruit (Passiflora edulis) vitamin A 25.0 Ripe banana (Musa sapientum)
9.7 Avocado (Persea Americana)
9.7 Pineaple (Ananas comosus)
Other vegetables not 89.8 43.9 Tomato (Solanum lycopersicum) rich in vitamin A 24.4 Onion (Allium cepa)
106
14.6 Green pepper (Capsicum annum)
2.4 Eggplant (solanum melongena)
Legumes and nuts 62.1 45.4 Groundnuts (Arachis hypogaes)
44.4 Beans (Vigna Angularis)
5.6 Soyabeans (Glycine max)
4.6 Cowpeas (Vigna unguiculata)
Meat, poultry, and fish 67.1 53.3 Beef
35.8 Fish
7.5 Chicken
Fats and oils 64.9 53.9 Ghee
23.1 Cooking oil
23.1 Margarine (Blue-band)
Dairy 24.5 76.9 Milk with tea
23.1 Whole milk
Eggs 15.1 84.2 Boiled eggs
These food groups were the basis of Dietary Diversity Score. In Food Variety Score all food items were included.
107
Table 5:4 The 24-hour dietary intake recall, recommended daily allowance, frequency of inadequate intake of nutrients among HIV positive and HIV negative adults in Kampala, Uganda (n=131)
24-Hour Intake RDA Deficit Median Nutrient Median Q25 Q75 Women Men (NAR) intake (%)
Energy, kcal 1732 1093 2345 2700 2000 75.5 73
Protein, g 40.5 29.0 62.1 56 46 83.8 60
Total fat, g 36.8 22.2 59.0 30 30 122.7 39
Carbohydrate, g 287 188 414 130 130 144.2 9
Dietary fiber, g 30.8 18.2 49.7 38 25 97.2 53
Protein, % energy 10.0 8.0 12.0 12.5 12.5 80.0 80
Fat, % energy 19.0 14.0 26.0 22.5 22.5 84.4 39
CHO, % energy 70.0 62.0 75.0 60 60 116.7 9
Calcium, mg 196 101 499 1000 1000 19.6 95
Magnesium, mg 177 104 315 260 220 68.9 64
Zinc, mg 5.2 3.1 7.6 7.0 4.9 92.9 56
Iron, mg 7.9 4.9 13.0 14 29 38.6 86
108
Vitamin A, μg 208 70 551 600 500 35.7 74
Thiamin, mg 0.8 0.5 1.2 1.2 1.1 66.7 70
Riboflavin, mg 0.9 0.5 1.3 1.3 1.1 72.7 66
Vitamin B6 2.2 1.3 3.1 1.3 1.3 169.2 22
Vitamin C, mg 91 49 178 45 45 202.8 22
Vitamin D, μg 0 0 1.0 5 5 0 98
Folate, μg 341 172 473 400 400 85.3 63
RDA = Recommended Daily Allowance, CHO = carbohydrate. NAR (Nutrient Adequacy Ratio) = actual intake/recommended intake. FAO = Food and Agriculture Organization, WHO = World Health Organization 2002. All NAR for micronutrients estimated using FAO/WHO requirements 2002 except vitamin D which was based on adequate intake (Atkinson and Ward 2001). DRFI = Dietary Reference Intake. NAR for energy, protein, carbohydrates, fat, and dietary fiber were estimated using DRFI.
109
Table 5:5 Percent of inadequate 24-hour dietary recall intake among HIV positive and HIV negative adults with tuberculosis in Kampala, Uganda
Characteristic HIV+ with TB (n=31) HIV- with TB (n=32)
Men Women Men Women
(n=10) (n=21) (n=18) (n=14)
Energy (kcal) n (%) 8 (80) 16 (76) 12 (67) 10 (71)
Protein (g) n (%) 6 (60) 14 (67) 10 (56) 10 (71)
Total fat (g) n (%) 3 (30) 14 (67) 5 (28) 6 (43)
Carbohydrate (g) n (%) 1 (10) 6 (29) 0 (0) 1 (7)
Dietary fiber (g) n (%) 6 (60) 11 (52) 9 (50) 6 (43)
Protein, % energy n (%) 8 (80) 13 (62) 16 (89) 10 (71)
Fat, % energy n (%) 4 (40) 16 (76) 10 (56) 9 (64)
Carbohydrate, % energy n (%) 4 (40) 4 (19) 5 (28) 3 (21)
Calcium (mg) n (%) 8 (80) 20 (95) 17 (94) 14 (100)
Magnesium (mg) n (%) 9 (90) 13 (62) 8 (44) 10 (71)
Zinc (mg) n (%) 6 (60) 11 (52) 10 (56) 9 (64)
Iron (mg) n (%) 8 (80) 19 (90) 14 (78) 14 (100)
110
Vitamin A (RE) n (%) 7 (70) 15 (71) 14 (82) 6 (43)
Vitamin D (μg) n (%) 9 (90) 20 (95) 18 (100) 14 (100)
Ascorbic acid (mg) n (%) 2 (20) 6 (30) 5 (28) 2 (14)
Folate (μg) n (%) 8 (80) 15 (71) 13 (72) 7 (50) ap-value <0.001, bp-value <0.05. HIV+ = HIV positive, HIV- = HIV negative, TB = tuberculosis.
111
Table 5:6 Percent of inadequate 24-hour dietary recall intake among HIV positive and HIV negative adults without tuberculosis in Kampala, Uganda
HIV- with no TB HIV+ with no TB Characteristic (n=38) (n=30)
Women Men Men Women
(n=17) (21) (n=16) (n=14)
Energy (kcal) n (%) 15 (88) 14 (67) 11 (69) 10 (71)
Protein (g) n (%) 8 (47) 14 (67) 6 (38) 10 (71)
Total fat (g) n (%) 5 (29) 8 (38) 2 (13) 8 (57)b
Carbohydrate (g) n (%) 0 (0) 4 (19) 0 (0) 0 (0)
Dietary fiber (g) n (%) 8 (47) 13 (62) 7 (44) 9 (64)
Protein (% of energy) n (%) 13 (76) 17 (81) 15 (94) 13 (93)
Fat (% of energy) n (%) 10 (59) 14 (67) 9 (56) 8 (57)
Carbohydrate (% of energy) n (%) 3 (18) 2 (10) 0 (0) 4 (29)b
Calcium (mg) n (%) 16 (94) 20 (95) 16 (100) 13 (93)
Magnesium (mg) n (%) 13 (76) 14 (67) 8 (50) 9 (64)
Zinc (mg) n (%) 10 (59) 11 (52) 7 (44) 10 (71)
112
Iron (mg) n (%) 14 (82) 19 (90) 11 (69) 14 (100)b
Vitamin A (RE) n (%) 15 (88) 12 (60) 14 (88) 12 (86)
Vitamin D (μg) n (%) 17 (100) 21 (100) 14 (100) 16 (100)
Ascorbic acid (mg) n (%) 4 (23) 3 (15) 3 (19) 2 (14)
Folate (μg) n (%) 11 (65) 14 (67) 4 (25) 10 (71)b ap-value <0.001, bp-value <0.05. HIV+ = HIV positive, HIV- = HIV negative, TB = tuberculosis.
113
Table 5:7 Spearman’s correlations between Nutrient Adequacy Ratio (NAR) of nutrients and Food Variety Score or Dietary Diversity Score among HIV positive and HIV negative adults with or without tuberculosis in Kampala, Uganda (n=131)
Women Men
(n=70) (n=61)
NAR Nutrient Food Dietary Food Dietary
Variety Diversity Variety Diversity
Score1 Score2 Score1 Score2
NAR Energy (kcal) 0.69 0.56 0.53 0.54
NAR Protein (g) 0.66 0.53 0.50 0.54
NAR Total fat (g) 0.59 0.53 0.54 0.57
NAR Carbohydrate (g) 0.59 0.49 0.46 0.38b
NAR Dietary fiber (g) 0.50 0.52 0.26b 0.37b
NAR Protein (% of energy) 0.18* 0.18* 0.11* 0.10*
NAR Fat (% of energy) 0.18* 0.21b 0.23* 0.26b
NAR Carbohydrate (% of energy) -0.22* -0.20* -0.28b -0.29b
NAR Calcium (mg) 0.70 0.69 0.22* 0.39
NAR Magnesium (mg) 0.63 0.51 0.43 0.31b
114
NAR Zinc (mg) 0.59 0.44 0.39b 0.34b
NAR Iron (mg) 0.59 0.49 0.29b 0.38b
NAR Vitamin A (RE) 0.50 0.59 0.30b 0.61
NAR Thiamin (mg) 0.59 0.51 0.27b 0.39b
NAR Riboflavin (mg) 0.61 0.55 0.35b 0.60
b NAR Vitamin B6 0.48 0.46 0.40 0.49
NAR Ascorbic acid (mg) 0.38b 0.42 0.36b 0.45
NAR Vitamin D (μg) 0.29b 0.44 0.08* 0.35b
NAR Folate (μg) 0.43 0.43 0.62 0.52
Mean Adequacy Ratio (MAR) 0.70 0.64 0.45 0.60
All correlation p-values <0.001 except bp-value < 0.05, *p-value >0.05. 1Number of all food items used in the study period. 2Number of all food groups consumed (1 – 9): 1) Cereal, roots, and tubers, 2) vitamin-A-rich fruits and vegetables, 3) 0ther fruits not rich in vitamin A, 4) other vegetables not rich in vitamin A, 5) legumes and nuts, 6) meat, poultry, and fish, 7) fats and oils, 8) diary, and 9) eggs and eggs products.
115
Table 5:8 Mean MAR scores for different levels of Food Variety Score (FVS) and
Dietary Diversity Score (DDS) among HIV positive and HIV negative adults in
Kampala, Uganda (n=131)
Dietary Food Variety Score
Diversity Score 1-3 4-6 7-9 10-12 13-15
Women (n=70)
1 24 (2) - - - -
2 26 (n=2) 46 (n=5) - - -
3 - 59 (n=4) - - -
4 - 39 (n=8) 65 (n=9) 63 (n=1) -
5 - 44 (n=3) 65 (n=12) 71 (n=5) 97 (n=1)
6 - - 73 (n=4) 87 (n=4) 90 (n=2)
7 - - 77 (n=3) 93 (n=2) 79 (n=1)
8 - - - 85 (n=1) 90 (n=1)
9 - - - - -
Men (n=61)
1 37 (n=2) - - - -
116
2 - 57 (n=1) - - -
3 - 65 (n=1) - - -
4 - 58 (n=7) 58 (n=9) 68 (n=2) -
5 - 73 (n=4) 73 (n=7) 75 (n=12) 90 (n=1)
6 - - 73 (n=5) 89 (n=7) -
7 - - 91 (n=1) 80 (n=1) -
8 - - - 92 (n=1) -
9 - - - - -
117
Table 5:9 Estimated Mean Adequacy Ratio scores for different levels of Food
Variety Score (FVS) and Dietary Diversity Score (DDS) from Linear Regression
Model among HIV positive and HIV negative adults in Kampala, Uganda (n=131)
Dietary Food Variety Score
Diversity
Score 3 5 6 7 8 9 10 12 15 (DDS)
Women (n=70)
1 32 40 44 49 53 57 61 70 82
2 35 43 47 51 56 60 64 73 85
3 38 46 50 54 59 63 67 75 88
4 41 49 53 57 62 66 70 78 91
5 43 52 56 60 64 69 73 81 94
6 46 55 59 63 67 72 76 84 98
7 49 58 62 66 70 74 79 87 100
8 52 61 65 69 73 77 82 90 103
9 55 63 68 72 76 80 84 93 105
Men (n=61)
118
1 33 36 37 39 40 41 43 46 50
2 41 43 45 46 48 49 51 54 58
3 48 51 53 54 56 57 58 61 66
4 56 59 60 62 63 65 66 69 74
5 64 67 68 70 71 73 74 77 81
6 72 75 76 77 79 80 82 85 89
7 82 82 84 85 87 88 90 93 97
8 87 90 92 93 94 96 97 100 105
9 95 98 99 101 102 104 105 108 112
Regression model for women: MAR = 16.349 + 2.906*DDS + 4.196*FVS. For men: MAR = 20.593 + 7.785*DDS + 1.453*FVS
119
Linear Regression Model to Estimate MAR Scores for Women
Dependent variable: MAR; Predictors: DDS and FVS
R2 Adjusted R2 Coefficient of variation p-value
0.502 0.487 26.138 <0.001
Unstandardized coefficients
β SE T p-value
Constant 16.349 6.211 2.63 0.011
DDS 2.906 2.071 1.40 0.165
FVS 4.196 1.106 3.80 0.0003
120
Linear Regression Model to Estimate MAR Scores for Men
Dependent variable: MAR; Predictors: DDS and FVS
R2 Adjusted R2 Coefficient of variation p-value
0.408 0.387 19.752 <0.001
Unstandardized coefficients
β SE T p-value
Constant 20.593 8.118 2.54 0.014
DDS 7.785 1.932 4.03 0.0002
FVS 1.453 0.834 1.74 0.870
121
Figure 5:1 Sensitivity (sens) and specificity (spec) for different cut-off points of Food
Variety Score (FVS) among women: Mean Adequacy Ratio (MAR) changing from
60 to 75
120
100
MARSe: 60 80 MARSe: 65 MARSe: 70 60 MARSe: 75 MARSp: 60 40 MARSp: 65 MARSp: 70 % % sensitivityand specificity 20 MARSp: 75
Cut-off point for FVS 0 1' 2' 3' 4' 5' 6' 7' 8' 9' 10' 11' 12' 13' 14' 15' MARSe: 60 0 7 7 14 31 48 69 76 79 97 100 100 100 100 100 MARSe: 65 0 5 5 11 24 41 62 73 76 92 97 100 100 100 100 MARSe: 70 0 5 5 10 22 37 56 66 76 93 98 100 100 100 100 MARSe: 75 0 5 5 9 21 35 53 65 77 93 98 100 100 100 100 MARSp: 60 100 100 100 100 100 98 90 78 63 41 29 17 12 5 2 MARSp: 65 100 100 100 100 100 100 97 88 70 45 33 21 15 6 3 MARSp: 70 100 100 100 100 100 100 97 86 76 52 38 24 17 7 3 MARSp: 75 100 100 100 100 100 100 96 89 81 56 41 26 19 7 4
Sensitivity MAR 60 – 75 = Sensitivity for a given cut-off point of FVS with a cut-off point for MAR varying from 60 to 75. Specificity MAR 60 – 75 = Sensitivity for a given cut-off point of FVS with a cut-off for MAR varying from 60 to 75. Sensitivity = identify nutritionally inadequate diets as inadequate; specificity = identify nutritionally adequate diets as adequate.
122
Figure 5:2 Sensitivity (sens) and specificity (spec) for different cut-off points of Food
Variety Score (FVS) among men: Mean Adequacy Ratio (MAR) changing from 60
to 75
120
100
MARSe: 60 80 MARSe: 65 MARSe: 70 60 MARSe: 75 MARSp: 60 40 MARSp: 65 MARSp: 70 % % sensitivityand specificity 20 MARSp: 75
Cut-off point 0 for FVS 1' 2' 3' 4' 5' 6' 7' 8' 9' 10' 11' 12' 13' 14' 15' MARSe: 60 0 0 6 13 19 31 38 44 69 75 88 88 100 100 100 MARSe: 65 0 0 4 8 12 28 44 48 76 80 88 92 100 100 100 MARSe: 70 0 0 3 7 13 27 40 43 70 83 90 93 100 100 100 MARSe: 75 0 0 3 6 15 26 38 41 68 82 88 94 100 100 100 MARSp: 60 100 100 100 100 96 91 80 76 56 42 31 16 2 0 0 MARSp: 65 100 100 100 100 94 94 89 83 67 50 36 19 3 0 0 MARSp: 70 100 100 100 100 97 97 90 84 68 58 42 23 3 0 0 MARSp: 75 100 100 100 100 100 100 93 85 70 63 44 26 4 0 0
Sensitivity MAR 60 – 75 = Sensitivity for a given cut-off point of FVS with a cut-off point for MAR varying from 60 to 75. Specificity MAR 60 – 75 = Sensitivity for a given cut-off point of FVS with a cut-off for MAR varying from 60 to 75. Sensitivity = identify nutritionally inadequate diets as inadequate; specificity = identify nutritionally adequate diets as adequate.
123
Figure 5:3 Sensitivity (sens) and specificity (spec) % for different cut-off points of
Diet Diversity Score (DDS) among women: Mean Adequacy Ratio (MAR) changing
from 60 to 75
120
100 MARSe: 60 80 MARSe: 65 MARSe: 70
60 MARSe: 75 MARSp: 60 MARSp: 65 % % sensitivityand specificity 40 MARSp: 70 MARSp: 75 20
Cut-off points for DDS 0 1' 2' 3' 4' 5' 6' 7' 8' 9' MARSe: 60 0 7 28 38 72 100 100 100 100 MARSe: 65 0 5 24 32 70 95 100 100 100 MARSe: 70 0 5 22 29 63 93 98 100 100 MARSe: 75 0 5 21 28 63 93 98 100 100 MARSp: 60 100 100 98 95 76 44 20 5 0 MARSp: 65 100 100 100 97 85 48 24 6 0 MARSp: 70 100 100 100 97 83 52 24 7 0 MARSp: 75 100 100 100 96 85 56 26 7 0
Sensitivity MAR 60 – 75 = Sensitivity for a given cut-off point of DDS with a cut-off point for MAR varying from 60 to 75. Specificity MAR 60 – 75 = Sensitivity for a given cut-off point of DDS with a cut-off for MAR varying from 60 to 75. Sensitivity = identify nutritionally inadequate diets as inadequate; specificity = identify nutritionally adequate diets as adequate.
124
Figure 5:4 Sensitivity (sens) and specificity (spec) % for different cut-off points of
Diet Diversity Score (DDS) among men: Mean Adequacy Ratio (MAR) changing
from 60 to 75
120
100 MARSe: 60 80 MARSe: 65 MARSe: 70
60 MARSe: 75 MARSp: 60 MARSp: 65 % % sensitivityand specificity 40 MARSp: 70 MARSp: 75 20
Cut-off points for DDS 0 1' 2' 3' 4' 5' 6' 7' 8' 9' MARSe: 60 0 0 19 19 69 100 100 100 100 MARSe: 65 0 0 12 16 64 96 100 100 100 MARSe: 70 0 0 10 13 63 97 100 100 100 MARSe: 75 0 0 9 12 59 91 100 100 100 MARSp: 60 100 100 100 98 76 33 7 2 0 MARSp: 65 100 100 100 100 83 39 8 3 0 MARSp: 70 100 100 100 100 90 45 10 3 0 MARSp: 75 100 100 100 100 93 44 11 4 0
Sensitivity MAR 60 – 75 = Sensitivity for a given cut-off point of DDS with a cut-off point for MAR varying from 60 to 75. Specificity MAR 60 – 75 = Sensitivity for a given cut-off point of DDS with a cut-off for MAR varying from 60 to 75. Sensitivity = identify nutritionally inadequate diets as inadequate; specificity = identify nutritionally adequate diets as adequate.
125
CHAPTER 6
PREDICTORS OF FAT MASS AND LEAN TISSUE AMONG HIV POSITIVE
AND HIV NEGATIVE ADULTS WITH OR WITHOUT TUBERCULOSIS IN
URBAN KAMPALA, UGANDA
126
Abstract
Background Fat and fat-free mass body composition measurements have been reported
to permit a more precise evaluation of body wasting and malnutrition than use of body
mass index; however, factors that may influence fat and fat-free mass and body mass index (BMI) have not been well characterized.
Objective We evaluated whether energy and protein intake are predictors of fat and fat- free mass and BMI in a population of HIV positive and HIV negative adults with or without tuberculosis in urban Kampala, Uganda.
Methods In a cross-sectional study with 131 adults who were screened for active tuberculosis and HIV infection, energy and protein intakes using a 24-hour dietary recall, weight, height, fat and fat-free mass using bioelectrical impedance analysis were assessed.
Results Energy intake was associated with an increase in BMI among women although intake in the presence of tuberculosis was associated with a decrease in BMI. Protein intake among women with no income and among women with unemployment was associated with a decrease in fat-free mass and fat mass, respectively whereas protein intake among women with tuberculosis was associated with an increase in BMI. Being single among women was associated with an increase in fat-free mass whereas having reduced appetite was associated with a decrease in fat-free mass and fat mass. Among men, tuberculosis was associated with a decrease in fat-free mass. Similarly, having reduced appetite was associated with decrease in fat mass and BMI. HIV did not influence body composition regardless of gender.
127
Conclusion The present study has revealed that there are remarkable gender differences in how energy and protein intakes influence body composition, and there are important interactions with presence of tuberculosis, lack of income, and unemployment. HIV does not appear to influence nutrient intake on body composition. Further evaluation is needed to understand how these differences would influence body composition over time and survival.
128
Background
Tuberculosis and human immunodeficiency virus (HIV) are diseases of major public
health concern worldwide, with the highest burden in sub-Saharan Africa (WHO 2009;
Lawn and Churchyard 2009). Among people with tuberculosis and HIV co-infection,
tuberculosis is the leading cause of death in sub-Saharan Africa (Lawn and Churchyard
2009; Corbett et al. 2003).
Both tuberculosis and HIV are independently associated with body wasting, and the
wasting in tuberculosis patients is thought to be further exacerbated by the concomitant
effects of HIV (Kotler 2000; Macallan 1999; Lucas et al. 1994). In contrast, findings from several cross-sectional studies (Niyongabo et al. 1999; Mupere et al. 2010; Paton and Ng 2006) appear to show no significant differences in body composition between
HIV positive adults with tuberculosis and HIV negative adults with tuberculosis suggesting that tuberculosis is the primary factor in driving the process wasting during co-infection. However, gender differences in body composition at presentation among tuberculosis patients have been reported (Kennedy et al. 1996; Mupere et al. 2010).
Although poorly understood, the wasting process in tuberculosis and HIV is probably due to a number of factors, including reduced nutrient intake, malabsorption, altered metabolism, and interactions between drugs and nutrients (Paton et al. 1999; Paton et al.
2003; Macallan et al. 1998; Fields-Gardner 1995; Kotler and Grunfeld 1995).
129
Fat and fat-free mass body composition measurements have been reported to permit a
more precise evaluation of body wasting than use of body mass index (BMI) (Kyle,
Piccoli, and Pichard 2003; VanItallie et al. 1990; Kyle et al. 2003). Evaluation of fat and
fat-free mass allows specific understanding of the body compartment and the extent to which the compartment is involved in the wasting. Moreover, disproportionate loss of fat-
free mass referred to as lean tissue may be associated with adverse effects on survival and
physical function.
BMI is insensitive to body fatness, particularly at low BMI, as well as with above-normal
muscle development (Kyle, Genton, and Pichard 2002; Kyle, Piccoli, and Pichard 2003).
The use of height-normalized fat and fat-free mass indexes (fat mass index (FMI) and fat-
free mass index (FFMI), respectively) allows partitioning of BMI into fat mass index
(FMI) and fat-free mass index (FFMI), i.e., BMI = FFMI + FMI (VanItallie et al. 1990;
Kyle et al. 2003). The height-normalized FMI and FFMI eliminate differences in fat and
fat-free mass associated with height (Baumgartner et al. 1998). Bioelectrical impedance
analysis (BIA) has been recommended as the preferred and precise method for clinical
assessment of FFM and fat mass (Kyle, Genton, and Pichard 2002; Kyle et al. 2004).
Prior studies conducted in African patients (Niyongabo et al. 1999; Villamor et al. 2006)
using BIA have reported reduced fat and fat-free mass at the time when patients present
with tuberculosis. However, the factors that may influence fat and fat-free mass indexes and BMI have not been well characterized.
130
In this cross-sectional study, we evaluated whether energy and protein intake are
predictors of fat and fat-free mass indexes and BMI in a population of HIV positive and
HIV negative with or without tuberculosis in urban Kampala, Uganda.
Methods
We enrolled 132 participants in a cross-sectional study with age 18 or more years residing in Kampala district or 20 km from the study site if residence was outside
Kampala in Uganda. Data collection was conducted between November 2007 and March
2008, a period that coincides with harvesting and light rains in November and December and dry season in January and February. One participant was excluded from the analysis because of prior tuberculosis treatment. The study was conducted at the National
Tuberculosis and Leprosy Program (NTLP) Clinic of the national tertiary teaching hospital, Mulago complex. Of the 131 participants who were included in the analysis, 31 were HIV positive and 32 HIV negative tuberculosis patients who were recruited at the
Mulago NTLP Clinic; 38 were HIV positive without tuberculosis and recruited at the
Infectious Disease Institute Clinic (IDI) located 500 meters from the Mulago NTLP
Clinic; and 30 were HIV negative adults without tuberculosis from the community where enrolled tuberculosis patients resided. The institutional review boards at Case Western
Reserve University and Joint Clinical Research Center approved the study, with final approval by the Uganda National Council for Science and Technology. All participants provided written informed consent to the study.
131
All subjects in the study were given appropriate pre- and post-test HIV counseling and
AIDS education. HIV-1 infection was diagnosed on the basis of a positive enzyme-linked immunosorbent assay for HIV-1 antibodies (Recombigen; Cambridge Biotech,
Cambridge, MA). At enrollment, basic demographic information and a medical history were collected, and a standardized physical examination was conducted by a medical officer. Active pulmonary TB was confirmed by sputum smear microscopy and culture.
Patients with active tuberculosis were treated with standard four-drug chemotherapy for tuberculosis per guidelines of the Ugandan Ministry of Health. Similarly, HIV positive patients eligible for antiretroviral therapy were started on treatment and cotrimoxazole prophylaxis at the IDI clinic.
Anthropometric measurements included height and weight. Body-mass index (BMI) was computed using the relationship of weight in kilograms divided by height in meters squared (kg/m2). Weight was taken using Hanson digital electronic bathroom weighing
scales to the nearest 100g. Height was measured to the nearest mm by calibrated standing
height using a stadiometer. All anthropometric measurement values were the mean of
duplicates. The single-frequency bioelectrical impedance analyzer (BIA Detroit, MI, RJL
Systems) performing at 50 kHz and 800 mA was used for BIA measures with detecting
electrodes placed on the wrist and ankle and signal introduction electrodes placed on the
first joint of the middle finger and behind the middle toe. Before performing
measurements on each subject, the BIA instrument was calibrated using the
manufacturer’s recalibration device. The resistance and reactance were based on
measures of a series circuit (Kotler et al. 1996). BIA measurements were performed in
132
duplicate for each subject. The analyzer was calibrated monthly. Fat-free mass was
calculated from BIA measurements using equations that were previously cross-validated
in a sample of patients (white, black and Hispanic) with and without HIV infection
(Kotler et al. 1996) and have been applied elsewhere in African studies (Shah et al. 2001;
Van Lettow et al. 2004; Villamor et al. 2006). Fat mass was calculated as body weight
minus fat-free mass.
The dietary intake assessment was made using a single 24-hour dietary recall with open
ended questions. The reference period for the 24-hour recall was the day prior to the day of the interview. In all instances, the interview was held only with the interviewee, no one else was present except for children. The questionnaire was pre-tested by administering it to 8 individuals selected randomly from the neighboring community to the study site. The assessment was conducted by four trained study nursing staffs and continuously supervised by a nutritionist using local food photographs, portion-size images, and
volumetric vessels to increase the accuracy of the recall from the previous 24 hours. The
nutritive value of raw ingredients was computed using the East African food composition
table database whose database was imported into the NutriSurvey software
(http://www.nutrisurvey.de) to easy the computations. The database used was
predominantly for local Ugandan diet. When the East African food composition table was
found deficient in certain food items, the United States Department of Agriculture
database and the African composition table were used.
133
Analytic strategy
Spearman correlations were performed between fat-free mass, fat mass, or BMI and energy or protein intake to establish the strengths of association. To establish whether energy and protein intakes are predictors of fat-free mass, fat mass, and BMI, we performed multivariable linear regression analyses with energy and protein intakes as main independent variables including all variables that were associated with a p < 0.50 in the unadjusted analyses (Dales and Ury 1978). Height-normalized indexes of fat-free mass, fat mass, and body mass were used in the models. The following variables were adjusted for in multivariable analyses: older age group >30 years, having tuberculosis and
HIV infection, no or low level (primary) of education, being single, being separated or divorced, having a household number of more than 2 people, unemployment, having no personal income, current history of alcohol intake, and reduced appetite.
Both univariate and multivariable analyses have been presented in this article. Previous
findings (Mupere et al. 2010) revealed gender differences in body composition among
individuals with or without tuberculosis regardless of HIV status. Further, there was
significant interaction between gender and energy intake. Thus, we stratified our analysis
according to gender. Two-way interactions were tested between energy or between
protein intakes for all variables involved in each multivariable model. The R-square was
used to evaluate the importance variables in the model. Influence diagnostics of the final
results was conducted to ensure that the results were not unduly influenced by a few
outlying measurements (Vittinghoff E. et al. 2005). The final regression coefficients were
134
obtained after eliminating participants with influential observations (Neter J., Wasserman
W., and Kutner M.H. 1990). Removal of influential points yielded more conservative
standard errors of the estimates. All analyses were performed using SAS version 9.2
(Cary software, North Carolina SAS Institute Inc 2004.)
Results
The characteristics of the study population are shown in Table 6:1 and 6:2. Of the 131
participants who were included in the analysis, 53% were females, 47% were males, 53%
were HIV positive, and 48% had tuberculosis.
The strengths of association between energy or protein intake and fat-free mass, fat mass,
or BMI are shown by spearman’s correlation coefficients (Table 6:3). Energy and protein
intakes had positive correlations with fat-free mass, fat mass and BMI regardless of gender. Among women, there were significant correlations between energy or protein intake and fat-free mass, fat mass, BMI, or with height-normalized indexes of fat and fat-
free mass.
Predictors of fat-free mass
In univariate analysis for the total study population, energy and protein intakes and being
single were associated with an increase in fat-free mass whereas tuberculosis, lack of any income, and having reduced appetite were associated with a decrease in fat-free mass
135
(Tables 6:4 and 6:5). Of note, energy intake among women was associated with a double
0.0004 ± 0.0002 standard error (SE) (p=0.028) increase in fat-free mass compared to men
that had 0.0002 ± 0.0003 SE, p=0.418 (Table 6:5). Having tuberculosis and reduced
appetite were associated with a decrease in fat-free mass regardless of gender.
In multivariable analysis for the total study population, there was significant interaction
between energy intake and gender when the model had gender, HIV status, tuberculosis
status, being single, being more two people per household, unemployment status, lack of
any income, and reduced appetite as adjusters. Energy intake among women was
associated with a significant decrease in fat-free mass, -0.001 ± 0.0003 SE, p=0.004). We thus performed stratified models according to gender adjusting for the same variables but without gender (Table 6:6 and 6:7). In the overall multivariable model, energy intake was associated with an increase in fat-free mass; however, this effect was not noticeable after stratification by gender (Table 6:6 and 6:7). Having tuberculosis was associated with a significant decrease in fat-free mass for the overall population; however, this decrease was double among men compared to women. Having reduced appetite was associated with a significant decrease in fat-free mass. There was a significant interaction between protein intake and income. Protein intake among women with no income was associated with a significant decrease in fat-free mass (Table 6:7). Of note, HIV was not associated with prediction of fat-free mass regardless of gender.
136
Predictors of fat mass
Energy intake in univariate analysis was associated with a decrease in fat mass of -
0.00002 ± 0.0003 SE, p=0.945 for the total population; however, this decrease was noted among men. Energy intake among women was associated with a significant increase in fat mass of 0.001 ± 0.0005 SE, p=0.020 (Table 6:8 and 6:9). Having tuberculosis and reduced appetite were associated with a decrease in fat mass regardless of gender. Of note, HIV was not associated with prediction of fat mass.
There was significant interaction between energy intake and gender in multivariable analysis for the overall study population when energy and protein intakes, gender, tuberculosis, being more two people per household, unemployment, lack of any income, and reduced appetite were independent variables in the model (Table 6:10 and 6:11).
Energy intake among women was associated with a significant increase in fat mass, 0.002
± 0.001, SE, p=0.001. Thus, in stratified analysis according to gender adjusting for the same variables, having reduced appetite was associated with a decrease in fat mass regardless of gender. Of note, women had a fourfold decrease (-4.17 ± 1.94, SE) in fat mass compared to men (-1.90 ± 0.90, SE) (Table 6:11). There was a significant interaction between protein intake and unemployment. Lack of employment among women was associated with a decrease in fat mass.
137
Predictors of BMI
In univariate analysis, energy intake was associated with an increase in BMI of 0.002 ±
0.001 SE, p=0.014 among women (Table 6:12 and 6:13). Energy intake among men was not significant. Having tuberculosis and reduced appetite were associated with a decrease in fat mass regardless of gender. Of note, HIV was not associated with prediction of fat mass (Table 6:12 and 6:13).
In multivariate analysis, after adjusting for protein intake, tuberculosis, lack of income, and reduced appetite, energy intake was associated an increase of 0.003 ± 0.001 ± SE among women (Table 6:14 and 6:15). Energy intake among men was not significant; however, having reduced appetite was associated with a four unit decrease in BMI among men. There were significant interactions between energy intake and tuberculosis, between energy intake and appetite, and between protein intake and tuberculosis (Table 6:14 and
6:15). Energy intake in the presence of tuberculosis among women was associated with a significant decrease in BMI whereas protein intake in the presence of tuberculosis was associated with a significant increase in BMI among women. None of these were significant among men. Energy intake with reduced appetite was associated with an increase in BMI in the overall population; however, this interaction not significant among women and among men (Table 6:14 and 6:15).
138
Discussion
In the present cross-sectional study of 131 participants, we evaluated whether energy and
protein intakes were important predictors of fat-free mass, fat mass, and BMI in a population of HIV positive and HIV negative adults with or without tuberculosis in urban
Uganda. Predictors of fat and fat-free mass and BMI differed by gender. Energy intake was associated with an increase in BMI among women although intake in the presence of tuberculosis was associated with a decrease in BMI. Protein intake among women with no income and among women with unemployment was associated with a decrease in fat- free mass and fat mass, respectively whereas protein intake among women with tuberculosis was associated with an increase in BMI. Being single among women was associated with an increase in fat-free mass whereas having reduced appetite was associated with a decrease in fat-free mass and fat mass. Among men, tuberculosis was associated with a decrease in fat-free mass. Similarly, having reduced appetite was associated with decrease in fat mass and BMI. HIV did not influence body composition regardless of gender.
The cross-sectional design employed in the present study limits the interpretation of finding in the present study to associations rather than causation. The findings suggest that energy and protein intakes, appetite, income, employment, and marital status are important predictors of body composition among women whereas tuberculosis and appetite are important among men. Energy intake in the presence of tuberculosis; reduced appetite among women and among men; tuberculosis among men; protein intake when
139
there is no income and when there is no employment were associated with negative prediction of body composition. Whereas protein intake when there is tuberculosis among women and being a single woman were associated with positive prediction of body composition. To our knowledge, this study is the first to show how energy and protein intakes influence body composition among women and among men and the interactions with tuberculosis, income, and employment. Of note, HIV infection did not provide any prediction of body composition regardless of gender. This corroborates with the claim in previous reports (Paton and Ng 2006; Mupere et al. 2010) that tuberculosis is the dominant factor in the wasting process even in the presence of HIV and that the disease in itself is associated with wasting (Rubin 1995). The major strengths in the present study is the composition of the study population with HIV positive and HIV negative adults with or without tuberculosis, allowing prediction of body composition in the presence or absence of tuberculosis and HIV infection.
This study found that the influence of energy and protein intakes on body composition differed by gender. A unit increase in energy intake was associated with a unit increase in
BMI among women; however, in the face of tuberculosis a unit increase in energy intake was associated with a unit decrease in BMI. Whereas protein intake among women with no income or unemployment were associated with a decrease in fat-free mass and fat mass, and protein intake in the presence of tuberculosis was associated with increase in
BMI. Energy and protein intakes among men did not influence body composition. A possible explanation to this gender difference in use of the dietary intake nutrient substrates rests on the differences in lipid and carbohydrate metabolism between men and
140
women (Tarnopolsky and Ruby 2001; Tarnopolsky 2000, 2008). Women use energy intake to build or maintain their fat content. Thus, an increase in BMI since BMI is a measure of fat content (Garrow and Webster 1985). However, during sub-maximal stress like lack of income or unemployment to purchase adequate nutrient substrates, they do oxidize away the lipid content sparing the lean tissue, and hence the decrease in BMI. It has reported in several reports that women oxidize more lipid and less of carbohydrate as metabolic substrates than men (Tarnopolsky 2000, 2008; Tarnopolsky and Ruby 2001).
However, in extreme stress such having tuberculosis disease, women may use protein substrates to maintain the BMI. For men, it appears they are in position to balance the use of nutrient substrates in building and maintaining both fat and fat-free mass content.
Thus, no effect is noticed in prediction models. Moreover, it has also been reported that at the time of tuberculosis diagnosis, the average weight group difference in men consisted of lean tissue and fat in equal proportions whereas in women, the average weight group difference consisted predominantly of fat mass (Mupere et al. 2010).
Findings in the present also show gender differences in the effect of reduced on body composition. Reduced appetite among women was associated with decrease in both fat and fat-free mass whereas among men, reduced appetite was associated with decrease in
BMI. This suggests that reduced appetite could be a predictor of both fat and fat-free mass among women whereas reduced appetite could be a predictor of BMI among men.
This probably reflects on the wasting process and the gender differences in body composition when anorexia sets in. Men waste away their high fat-free mass content to be in proportion with fat content whereas women, waste away their high fat content
141
sparing the low fat-free mass content. Although, BMI is monotonically related to
adiposity (Garrow and Webster 1985), it also correlates positively with the amount of fat-
free an individual has. Hence, the level of appetite may reflect on the body composition
for an individual. Women have little fat-free mass content, thus requiring daily energy intake to preserve physical function. Men in general display a higher absolute resting metabolic rate with associated energy expenditure than women because of their lager quantity of fat-free mass (Arciero, Goran, and Poehlman 1993). Thus, men require a good
appetite to maintain nutrient intake that can meet their metabolic demands.
Another possible explanation is the differences in sex hormones between men and
women that the sex hormones may play a role in regulation of symptoms associated
anorexia in tuberculosis. Sex hormones are physiologically associated with feeding
behavior (Geary 2001). Higher anorectic signals and earlier satiety have been reported in
men suffering from chronic illnesses, perhaps contributing to a different response pattern
to anorexigenic diseases such as cancer among men and women (Geary 2001). In animal
models, inflammation-induced anorexia has been reported to be more severe among male
rats (Lennie 2004) and previous reports suggest that estradiol and progesterone have
inhibitory effects on anorexia (Eckel 2004). Further, sex hormones or derivatives have
been used in treatment of eating disorders with positive results in improving appetite,
caloric intake, and nutritional and inflammatory status (Rammohan et al. 2005).
142
Findings in the present revealed that tuberculosis was a universal predictor of body composition. Having tuberculosis was associated a decrease in fat and fat-free mass and
BMI regardless of gender. Wasting is a recognized cardinal feature of tuberculosis. It is likely caused by a combination of reduction in appetite, leading to a decrease in energy intake, interacting with increased losses and altered metabolism as part of the inflammatory and immune responses (Paton et al. 2003; Macallan et al. 1998). Utilization of amino acids and protein synthesis may be inhibited due to the presence of pro- inflammatory cytokines (Macallan et al. 1998).
In conclusion, the present study has revealed that there are remarkable gender differences in how energy and protein intakes influence body composition, and there are important interactions with presence of tuberculosis, lack of income, and unemployment. HIV does not appear to influence nutrient intake on body composition. Further evaluation is needed to understand how these differences would influence body composition over time and survival.
Acknowledgements
We thank all study staff members of the Case Western Reserve University and Makerere
University research collaboration at the Tuberculosis Research Unit in the United States and in Uganda for their assistance; the faculty and staff at Case Western Reserve
University Department of Epidemiology and Biostatistics for the guidance in analyzing the project; and the Fogarty International Center, for the continued support.
143
This study was supported in part by the AIDS International Training Research Program,
Fogarty International Center, Grant No. TW000011, based at Case Western Reserve
University, Department of Epidemiology and Biostatistics. This work was part of Ezekiel
Mupere’s PhD thesis at Case Western Reserve University.
144
Table 6:1 Characteristics of the study population (n=131)
Characteristics All subjects (n=131)
Sex
Female n (%) 70 (53)
Male n (%) 61 (47)
Age in years
≤30 n (%) 94 (72)
>30 n (%) 37 (28)
HIV status
Negative n (%) 62 (47)
Positive n (%) 69 (53)
Tuberculosis
No n (%) 68 (52)
Yes n (%) 63 (48)
Education
None/primary level n (%) 69 (53)
Secondary level n (%) 62 (47)
145
Tribe
Muganda n (%) 57 (44)
Others n (%) 74 (56)
Marital status
Married n (%) 56 (43)
Single n (%) 37 (28)
Separated/divorced n (%) 38 (29)
Household number
One to two n (%) 40 (31)
>2 n (%) 91 (69)
Employed
No n (%) 54 (41)
Yes n (%) 77 (59)
Income
Not at all n (%) 47 (36)
Yes n (%) 84 (64)
Takes alcohol
146
No n (%) 98 (75)
Yes n (%) 33 (25)
Reduced appetite
No n (%) 89 (68)
Yes n (%) 42 (32)
147
Table 6:2 Nutrient intake characteristics of the study population (n=131)
Characteristics All subjects (n=131)
Energy intake in kcal [mean, SD)] 1820 (874)
Protein intake in g [mean, (SD)] 49.5 (32.1)
Body mass index in kg/m2 [mean, (SD)] 20.9 (3.8)
Fat-free mass in kg [mean, (SD)] 45.0 (7.0)
Fat-free mass index in kg/m2 [mean, (SD)] 16.8 (1.6)
Fat mass in kg [mean, (SD)] 10.4 (7.0)
Fat mass index in kg/m2 [mean, (SD)] 4.1 (3.1)
SD = standard deviation
148
Table 6:3 Spearman’s correlations between energy or protein intake and body mass index, fat or fat-free mass (n=131)
Women (n=70) Men (n=61) Characteristics Energy (kcal) Protein (g) Energy (kcal) Protein (g)
Body mass index (kg/m2) 0.28b 0.08 0.10 0.12
Fat-free mass (kg) 0.36b 0.13 0.20 0.18
Fat-free mass index (kg/m2) 0.28b 0.06 0.11 0.11
Fat mass (kg) 0.29b 0.10 0.04 0.11
Fat mass index (kg/m2) 0.25b 0.08 0.02 0.09 ap-value <0.001, bp-value <0.05.
149
Table 6:4 Predictors of fat-free mass in univariate models among HIV positive and
HIV negative adults with or without tuberculosis in Kampala, Uganda (n=131)
All (n=131)
Characteristics Unadjusted p-value
Estimate (SE)
Energy intake, kcal 0.001 (0.0002) <0.001
Protein intake, g 0.002 (0.001) 0.049
Older age >30 years -0.23 (0.30) 0.449
HIV positive -0.34 (0.27) 0.210
Tuberculosis -1.41 (0.24) <0.001
No/low education -0.41 (0.43) 0.339
Not Muganda by tribe -0.34 (0.27) 0.148
Single 0.84 (0.29) 0.005
Household >2 people -0.32 (0.30) 0.279
Unemployed -0.46 (0.28) 0.095
No income at all -0.57 (0.28) 0.045
Takes alcohol -0.03 (0.32) 0.928
150
Reduced appetite -1.55 (0.26) <0.001
151
Table 6:5 Predictors of fat-free mass in univariate models among HIV positive and
HIV negative adults with or without tuberculosis in Kampala, Uganda stratified by sex (n=131)
Women (n=70) Men (n=61)
Characteristics Unadjusted p- Unadjusted p-
Estimate (SE) value Estimate (SE) value
Energy intake, kcal 0.0004 (0.0002) 0.028 0.0002 (0.0003) 0.418
Protein intake, g 0.001 (0.001) 0.275 0.0002 (0.002) 0.910
Older age >30 years -0.02 (0.32) 0.937 -0.65 (0.45) 0.157
HIV positive -0.35 (0.28) 0.212 0.14 (0.42) 0.743
Tuberculosis -0.960 (0.25) 0.0003 -1.81 (0.35) <0.001
No/low education 0.12 (0.38) 0.754 -0.60 (0.84) 0.481
Not Muganda by tribe -0.41 (0.27) 0.144 -0.44 (0.42) 0.301
Single 0.16 (0.28) 0.573 0.98 (0.58) 0.096
Household >2 people 0.10 (0.32) 0.746 -0.28 (0.43) 0.525
Unemployed 0.11 (0.28) 0.692 -0.49 (0.46) 0.283
No income at all 0.18 (0.27) 0.508 -0.62 (0.52) 0.236
Takes alcohol 0.25 (0.31) 0.416 -0.21 (0.50) 0.682
152
Reduced appetite -1.15 (0.25) <0.001 -1.64 (0.44) 0.0004
153
Table 6:6 Predictors of fat-free mass in multivariable models among HIV positive and HIV negative adults with or without tuberculosis in urban Kampala, Uganda
(n=131)
Overall (n=131) Characteristic Estimate (SE) p-value
Energy intake, kcal 0.0004 (0.0002) 0.025
Protein intake, kcal 0.0004 (0.001) 0.756
HIV positive -0.40 (0.24) 0.096
Tuberculosis -0.86 (0.29) 0.004
Single 0.68 (0.24) 0.005
Household >2 people -0.39 (0.26) 0.138
No income 0.36 (0.43) 0.408
Reduced appetite -0.88 (0.31) 0.006
Protein*income -0.02 (0.01) 0.036
R Square 0.41 ap-value <0.001, bp-value <0.05. Variables for which p-value was < 0.50 in the unadjusted analyses were included in the multivariable model. Unemployment and other than Muganda tribe were dropped because they did not contribute much to the R-square.
154
Table 6:7 Predictors of fat-free mass in multivariable models among HIV positive and HIV negative adults with or without tuberculosis in Kampala, Uganda stratified according sex (n=131)
Stratified models
Characteristic Women (n=70) Men (n=61)
Estimate (SE) p-value Estimate (SE) p-value
Energy intake, kcal 0.0003 (0.0002) 0.094 0.0004 (0.0004) 0.313
Protein intake, kcal 0.002 (0.001) 0.289 -0.002 (0.003) 0.383
HIV positive -0.35 (0.25) 0.168 -0.26 (0.42) 0.546
Tuberculosis -0.59 (0.33) 0.078 -1.42 (0.45) 0.003
Single 0.58 (0.25) 0.022 0.47 (0.44) 0.295
Household >2 people -0.36 (0.29) 0.227 0.14 (0.45) 0.755
No income 0.80 (0.41) 0.056 -0.16 (0.89) 0.861
Reduced appetite -0.79 (0.33) 0.020 -0.77 (0.52) 0.144
Protein*income -0.02 (0.01) 0.027 -0.007 (0.01) 0.591
R Square 0.42 0.40 ap-value <0.001, bp-value <0.05. Variables for which p-value was < 0.50 in the unadjusted analyses were included in the multivariable model. Unemployment and other than Muganda tribe were dropped because they did not contribute much to the R-square.
155
Table 6:8 Predictors of fat mass in univariate models among HIV positive and HIV negative adults with or without tuberculosis in Kampala, Uganda (n=131)
All (n=131) Characteristics Unadjusted Estimate (SE) p-value
Energy intake, kcal -0.00002 (0.0003) 0.945
Protein intake, g -0.001 (0.002) 0.536
Older age >30 years 0.09 (0.060) 0.885
HIV positive 0.49 (0.54) 0.367
Tuberculosis -2.40 (0.50) <0.001
No/low education 0.11 (0.85) 0.90
Not Muganda by tribe -0.30 (0.55) 0.590
Single -1.10 (0.59) 0.064
Household >2 people 0.82 (0.59) 0.162
Unemployed 0.88 (0.55) 0.112
No income at all 1.46 (0.55) 0.009
Takes alcohol 0.75 (0.62) 0.23
Reduced appetite -2.02 (0.55) <0.001
156
Table 6:9 Predictors of fat mass in univariate models among HIV positive and HIV negative adults with or without tuberculosis in Kampala, Uganda stratified according to sex (n=131)
Characteristics Women (n=70) Men (n=61)
Unadjusted p-value Unadjusted p-value
Estimate (SE) Estimate (SE)
Energy intake, kcal 0.001 (0.0005) 0.020 -0.000004 (0.0002) 0.981
Protein intake, g 0.004 (0.003) 0.304 0.0003 (0.001) 0.821
Older age >30 years 0.53 (0.95) 0.581 0.07 (0.32) 0.829
HIV positive -0.43 (0.85) 0.619 0.45 (0.29) 0.132
Tuberculosis -3.57 (0.72) <0.001 -1.35 (0.24) <0.001
No/low education -0.76 (1.15) 0.508 -0.29 (0.60) 0.632
Not Muganda by tribe -0.29 (0.84) 0.728 -0.20 (0.30) 0.508
Single -0.11 (0.85) 0.900 -0.006 (0.42) 0.988
Household >2 people 0.62 (0.97) 0.524 0.06 (0.31) 0.849
Unemployed 0.48 (0.83) 0.563 -0.28 (0.33) 0.390
No income at all 0.75 (0.83) 0.374 -0.06 (0.38) 0.869
Takes alcohol 1.19 (0.93) 0.203 -0.23 (0.35) 0.519
157
Reduced appetite -3.47 (0.75) <0.001 -1.30 (0.30) <0.001
158
Table 6:10 Predictors of fat mass in multivariable models among HIV positive and
HIV negative adults with or without tuberculosis in Kampala, Uganda (n=131)
Overall model (n=131) Characteristic Estimate (SE) p-value
Energy intake, kcal -0.0005 (0.0008) 0.534
Protein intake, g -0.009 (0.005) 0.077
Tuberculosis -0.33 (1.31) 0.799
Household >2 people -2.06 (1.11) 0.066
Unemployed 1.66 (1.17) 0.158
No income -0.80 (1.50) 0.596
Reduced appetite -3.99 (1.29) 0.003
Energy*household >2 people 0.002 (0.001) 0.010
Energy*tuberculosis -0.003 (0.001) <0.001
Energy*reduced appetite 0.002 (0.001) 0.005
Energy*no income 0.002 (0.001) 0.021
Protein*tuberculosis -0.09 (0.02) <0.001
Protein*unemployement -0.05 (0.02) 0.003
159
R Square 0.41 ap-value <0.001, bp-value <0.05. Variables for which p-vale was < 0.50 in the unadjusted analyses were included in the multivariable model. Alcohol intake and being single were dropped because they did not contribute much to the R-square.
160
Table 6:11 Predictors of fat mass in multivariable models among HIV positive and
HIV negative adults with or without tuberculosis in Kampala, Uganda (n=131)
Stratified models
Women (n=70) Men (n=61) Characteristic p- Estimate (SE) p- Estimate (SE) value value
Energy intake, kcal -0.0002 (0.002) 0.880 -0.001 (0.0003) 0.150
Protein intake, g -0.01 (0.01) 0.539 0.001 (0.003) 0.710
Tuberculosis -0.37 (2.15) 0.866 -1.49 (0.89) 0.100
Household >2 people -2.86 (1.72) 0.102 0.33 (0.90) 0.715
Unemployed 2.67 (1.85) 0.155 -0.57 (0.78) 0.468
No income -1.95 (2.40) 0.420 -0.84 (1.00) 0.407
Reduced appetite -4.17 (1.94) 0.036 -1.90 (0.90) 0.040
Energy*household >2 people 0.002 (0.001) 0.080 0.0001 (0.0004) 0.885
Energy*tuberculosis -0.003 (0.001) 0.070 0.0002 (0.001) 0.751
Energy*reduced appetite 0.001 (0.001) 0.207 0.001 (0.0004) 0.134
Energy*no income 0.002 (0.001) 0.097 0.001 (0.005) 0.264
161
Protein*tuberculosis 0.07 (0.04) 0.100 -0.003 (0.02) 0.859
Protein*unemployement -0.08 (0.03) 0.010 -0.001 (0.01) 0.938
R Square 0.53 0.51 ap-value <0.001, bp-value <0.05. Variables for which p-vale was < 0.50 in the unadjusted analyses were included in the multivariable model. Alcohol intake and being single were dropped because they did not contribute much to the R-square.
162
Table 6:12 Predictors of body mass index in univariate models among HIV positive and HIV negative adults with or without tuberculosis in Uganda (n=131)
Characteristics All (n=131)
Unadjusted p-value
Estimate (SE)
Energy intake, kcal 0.001 (0.0004) 0.161
Protein intake, g 0.001 (0.003) 0.786
Older age >30 years -0.10 (0.74) 0.894
HIV positive 0.15 (0.66) 0.823
Tuberculosis -3.88 (0.57) <0.001
No/low education -0.27 (1.04) 0.797
Not Muganda by tribe -0.71 (0.67) 0.292
Single -0.22 (0.73) 0.762
Household >2 people 0.53 (0.72) 0.460
Unemployed 0.36 (0.67) 0.594
No income at all 0.89 (0.69) 0.198
Takes alcohol -0.76 (0.76)) 0.319
163
Reduced appetite -3.67 (0.63) <0.001
164
Table 6:13 Predictors of body mass index in univariate models among HIV positive and HIV negative adults with or without tuberculosis in Kampala, Uganda stratified according sex (n=131)
Women (n=70) Men (n=61)
Characteristics Unadjusted p-value Unadjusted p-value
Estimate (SE) Estimate (SE)
Energy intake, kcal 0.002 (0.001) 0.014 0.0002 (0.0004) 0.622
Protein intake, g 0.005 (0.004) 0.306 0.0005 (0.003) 0.864
Older age >30 years 0.59 (1.22) 0.633 -0.57 (0.72) 0.434
HIV positive -0.76 (1.09) 0.490 0.59 (0.67) 0.385
Tuberculosis -4.66 (0.91) <0.001 -3.15 (0.54) <0.001
No/low education -0.57 (1.47) 0.698 -0.89 (1.35) 0.514
Not Muganda by tribe -0.72 (1.08) 0.505 -0.64 (0.67) 0.347
Single 0.08 (1.09) 0.945 0.97 (0.94) 0.306
Household >2 people 0.82 (1.24) 0.514 -0.22 (0.69) 0.755
Unemployed 0.52 (1.07) 0.625 -78 (0.73) 0.293
No income at all 0.96 (1.06) 0.372 -0.69 (0.84) 0.418
Takes alcohol 1.53 (1.19) 0.202 -0.44 (0.80) 0.587
165
Reduced appetite -4.78 (0.934) <0.001 -2.95 (0.68) <0.001
166
Table 6:14 Predictors of body mass index in multivariable models among HIV positive and HIV negative adults with or without tuberculosis in urban Uganda
(n=131)
Characteristic Overall model (n=131)
Estimate (SE) p-value
Energy intake, kcal 0.001 (0.001) 0.193
Protein intake, g -0.01 (0.01) 0.105
Tuberculosis -1.07 (1.52) 0.482
No income 1.01 (0.57) 0.079
Reduced appetite -4.93 (1.54) 0.002
Energy*tuberculosis -0.002 (0.001) 0.013
Energy*appetite 0.002 (0.001) 0.026
Protein*tuberculosis 0.06 (0.03) 0.033
R Square 0.39 ap-value <0.001, bp-value <0.05. Variables for which p-vale was < 0.50 in the unadjusted analyses were included in the multivariable model. Being other than Muganda tribe was dropped because it did not contribute much to the R-square.
167
Table 6:15 Predictors of body mass index in multivariable models among HIV positive and HIV negative adults with or without tuberculosis in Kampala, Uganda
(n=131)
Stratified models
Characteristic Women (n=70) Men (n=61)
Estimate (SE) p-value Estimate (SE) p-value
Energy intake, kcal 0.003 (0.001) 0.028 -0.001 (0.001) 0.311
Protein intake, g -0.02 (0.01) 0.119 0.003 (0.01) 0.622
Tuberculosis -0.48 (2.27) 0.834 -3.34 (1.76) 0.063
No income 1.01 (0.87) 0.250 -0.69 (0.67) 0.306
Reduced appetite -4.21 (2.25) 0.066 -4.3 (1.94) 0.030
Energy*tuberculosis -0.004 (0.001) 0.008 0.001 (0.001) 0.238
Energy*appetite 0.001 (0.001) 0.580 0.001 (0.001) 0.105
Protein*tuberculosis 0.10 (0.05) 0.039 -0.03 (0.03) 0.236
R Square 0.48 0.47 ap-value <0.001, bp-value <0.05. Variables for which p-vale was < 0.50 in the unadjusted analyses were included in the multivariable model. Being other than Muganda tribe was dropped because it did not contribute much to the R-square.
168
CHAPTER 7
BODY WASTING AND DIETARY INTAKE AMONG HIV POSITIVE AND HIV
NEGATIVE ADULTS WITH OR WITHOUT TUBERCULOSIS IN URBAN
UGANDA, KAMPALA
169
Abstract
Background The effects of tuberculosis and effect of HIV on dietary intake have not been well described. We aimed to establish the 1) independent effects of tuberculosis and
HIV infection on dietary intake, 2) relationship between dietary intake and body wasting
as measured by height-normalized fat-free mass (FFMI) and body mass (BMI) indices,
and 3) relationship between dietary intake and tuberculosis disease severity among HIV
positive and HIV negative adults with/or without tuberculosis in urban Uganda.
Methods In a cross-sectional study of 131 adults who were screened for active
tuberculosis and HIV infection, FFMI, BMI, and 24-hour dietary intake recall were
assessed.
Results Tuberculosis patients that had moderate/or severe clinical disease had lower
dietary intakes for energy, protein, total fat, carbohydrate, calcium, vitamin A, and folate
compared to patients with mild disease. Both men and women had comparable dietary
intake among patients with tuberculosis regardless of HIV status whereas HIV negative
women had reduced energy, protein, and folate intake among individuals without
tuberculosis compared to men. Tuberculosis patients with fat-free mass wasting or those
with reduced BMI had comparable nutrient intakes with counterparts that had normal fat- free mass or normal BMI.
Conclusion The study revealed that dietary intake at the time of diagnosis was influenced by tuberculosis disease severity, but not tuberculosis disease or HIV status and in the absence of tuberculosis was influenced by gender. Nutritional counseling and
170
supplementation, early treatment and prevention of tuberculosis are needed to improve dietary intake in populations of sub-Saharan Africa.
171
Background
Body wasting and malnutrition is endemic in sub-Saharan Africa (Muller and Krawinkel
2005), which also bears the highest burden of tuberculosis patients with human
immunodeficiency virus (HIV) co-infection (Lawn and Churchyard 2009). An estimated
1.37 million new cases of tuberculosis with HIV co-infection occurred in 2007; 79% of which were from sub-Saharan Africa. Co-infection with tuberculosis and HIV poses an extra burden to the pathophysiology of body wasting, exacerbating the wasting process seen in tuberculosis or HIV infection alone (Lucas et al. 1994; Macallan 1999).
Moreover, co-infection and malnutrition have deleterious interactions. Co-infected patients with malnutrition have high risk of morbidity and mortality (Zachariah et al.
2002; Lucas et al. 1994; Duarte et al. 2009) and tuberculosis is the leading cause of death in co-infected patients in tuberculosis endemic countries, including those with free access to antiretroviral therapy (Saraceni et al. 2008).
Co-infection may lead to poor appetite with decreased nutrient intake, which may interact with altered metabolism associated with both infections as part of the immune and inflammatory responses (Paton et al. 2003) leading to exacerbation of the existing body wasting. Yet poor nutritional status is associated with risk of tuberculosis relapse (Khan et al. 2006) in addition to morbidity and mortality. The goal of nutritional assessment and nutritional support is to intervene early and to preserve lean tissue or fat-free mass body compartment from further wasting because disproportionate loss of fat-free mass is associated with morbidity and mortality (Heitmann et al. 2000). Fat-free mass is a marker
172
of body wasting and malnutrition, because it is a consequence of negative imbalance
between energy (and protein) needs and dietary intake that occurs for more than a few
days.
Assessment of dietary intake is essential in nutritional management and in understanding
of body wasting and malnutrition. Despite the high burden of malnutrition and the high
burden of co-infection with associated body wasting, assessment of dietary intake is often
neglected in clinical practice and national tuberculosis programs. Thus, the effects of
tuberculosis and effect of HIV among co-infected patients on dietary intake have not been well described. The present cross-sectional study was conducted to establish the 1) independent effects of tuberculosis and HIV infection on dietary intake, 2) relationship between dietary intake and body wasting, and 3) relationship between dietary intake and
TB disease severity. The study was conducted among HIV positive and HIV negative adults with/or without active tuberculosis in urban Kampala, Uganda.
Subjects and methods
In a cross-sectional study, 132 participants 18 years or older residing in Kampala district or 20 km from the study site if residence was outside Kampala in Uganda were enrolled.
One participant was excluded from the analysis because of prior TB treatment. The study was conducted at the National tuberculosis and Leprosy Program (NTLP) Clinic of the national tertiary teaching hospital, Mulago complex between November 2007 and March
2008. Of the 131 participants who were included in the analysis, 63 were tuberculosis
173
patients who were recruited at the Mulago NTLP Clinic; 38 were HIV positive without
tuberculosis and recruited at the Infectious Disease Institute Clinic (IDI) located 500
meters from the Mulago NTLP Clinic; and 30 were HIV negative without tuberculosis
from the community where enrolled tuberculosis patients resided. The institutional
review boards at Case Western Reserve University and Joint Clinical Research Center
approved the study, with final approval by the Uganda National Council for Science and
Technology. All participants provided written informed consent to the study.
All subjects in the study were given appropriate pre- and post-test HIV counseling and
AIDS education. HIV-1 infection was diagnosed on the basis of a positive enzyme-linked immunosorbent assay for HIV-1 antibodies (Recombigen; Cambridge Biotech,
Cambridge, MA). At enrollment, basic demographic information and a medical history were collected, and a standardized physical examination was conducted by a medical officer. Active pulmonary tuberculosis was confirmed by sputum smear microscopy and culture. Patients with active tuberculosis were treated with standard four-drug chemotherapy for tuberculosis per guidelines of the Ugandan Ministry of Health.
Similarly, HIV positive patients eligible for antiretroviral therapy were started on treatment and cotrimoxazole prophylaxis at the IDI clinic.
We defined body wasting of participants using body mass index (BMI) and height- normalized indices (adjusted for height2) of body composition that partition BMI into fat-
free mass index (FFMI) and fat mass index (FMI) (Kyle, Piccoli, and Pichard 2003;
174
VanItallie et al. 1990; Schutz, Kyle, and Pichard 2002) to establish the body wasting status of participants. The FFMI and FMI have the advantages of compensating for differences in height and age (Kyle, Genton, and Pichard 2002). Also, the use of the
FFMI and FMI eliminates some of the differences between population groups. We defined body wasting as patients having the low fat-free mass index (FFMI) and the low body fat mass index (FMI) as previously recommended (Kyle, Piccoli, and Pichard 2003;
VanItallie et al. 1990). The low FFMI of 16.7 (kg/m2) for men and 14.6 (kg/m2) for women and the low FMI of 1.8 (kg/m2) for men and 3.9 (kg/m2) for women corresponds to a BMI of <18.5 kg/m2, the WHO cutoff for malnutrition (World Health Organ Tech
Rep 1995) among adults.
Anthropometric measurements included height and weight. Body-mass index (BMI) was computed using the relationship of weight in kilograms divided by height in meters squared (kg/m2). Weight was taken using Hanson digital electronic scales to the nearest
100g. Height was measured to the nearest cm by calibrated standing height using a stadiometer. All anthropometric measurement values were the mean of duplicates. The single-frequency bioelectrical impedance analyzer (BIA Detroit, MI, RJL Systems) performing at 50 kHz and 800 mA was used for BIA measures with detecting electrodes placed on the wrist and ankle and signal introduction electrodes placed on the first joint of the middle finger and behind the middle toe.
175
Before performing measurements on each subject, the BIA instrument was calibrated using the manufacturer’s recalibration device. The resistance and reactance were based on measures of a series circuit (Kotler et al. 1996). BIA measurements were performed in duplicate for each subject. The analyzer was calibrated monthly. Fat-free mass was calculated from BIA measurements using equations that were previously cross-validated in a sample of patients (white, black and Hispanic) with and without HIV infection
(Kotler et al. 1996) and have been applied elsewhere in African studies (Shah et al. 2001;
Van Lettow et al. 2004; Villamor et al. 2006). Fat mass was calculated as body weight minus fat-free mass.
We classified tuberculosis patients into two categories of clinical disease severity as follows: mild and moderate/or severe tuberculosis using the TBscore. HIV negative participants without tuberculosis were classified as no disease category. The clinical tuberculosis score is a low-cost tool that has been developed recently and validated to assess severity of clinical TB disease in resource limited settings (Wejse et al. 2008). The score has a maximum of 13 clinical variables each scoring 1 point as follows: self- reported symptoms of cough, hemoptysis, dyspnea, chest pain, and night sweating; anemic conjunctivae, tachycardia, positive finding at lung auscultation, axillary temperature >37.00C, body mass index (BMI) <18, BMI <16, middle upper arm circumference (MUAC) <220 mm, MUAC <200. Clinical severity categories were generated using the following cutoffs: mild 0 – 5 and moderate/severe > 5. Higher scores are associated with severe disease.
176
The dietary intake assessment was made using a single 24-hour dietary recall with open
ended questions. The questionnaire was pre-tested by administering it to 8 individuals selected randomly from the neighboring community to the study site. The assessment was conducted by four trained study nursing staffs and supervised a nutritionist using local food photographs, portion-size images, and volumetric vessels to increase the accuracy of the recall from the previous 24 hours. The nutritive value of raw ingredients was computed using the East African food composition table database whose database was imported into the NutriSurvey software (http://www.nutrisurvey.de) to easy the computations. When the East African food composition table was found deficient in certain food items, the United States Department of Agriculture database and the African composition table were used.
Analysis
All study participants in the analysis were categorized into 4 mutually independents groups: HIV positive patients with and without tuberculosis disease, HIV negative
patients with and without tuberculosis. Measures of central tendency and variability were
compared between women and men across 4 mutually exclusive groups, between patients
with and patients without reduced FFMI, between patients with and without reduced
BMI, and between patients with mild and with moderate/severe disease using Wilcoxon-
Mann Whitney test for average weight, height, BMI, fat and fat-free mass, FFMI, FMI,
and nutrient intake parameters due to lack of normality.
177
We made comparisons using analysis of variance between HIV positive patients with
tuberculosis and HIV positive patients without tuberculosis, between HIV negative
patients with tuberculosis and HIV negative individuals without tuberculosis, between
HIV positive patients with tuberculosis and HIV negative patients with tuberculosis, and
between HIV positive individuals without tuberculosis and HIV negative individuals
without tuberculosis to understand the independent effects of tuberculosis and HIV
infection on dietary intake and nutritional status. We used Bonferroni to adjust for
multiple comparisons. We compared proportions of nutrient inadequacy between women
and men across the 4 mutually exclusive groups, between patients with and without
reduced FFMI, between patients with and without reduced BMI, and between patients
with mild and patients with moderate/severe clinical tuberculosis using chi-square and
Fischer’s exact tests. Fischer’s exact test was used where expected counts were less than
5. A p-value of <0.05 was considered significant in all analyses except for multiple comparisons which was set at <0.01. All analyses were performed using SAS version 9.2
(Cary software, North Carolina SAS Institute Inc 2004).
Results
Of the 131 participants analyzed, 31 were HIV positive with tuberculosis, 32 were HIV negative with tuberculosis, 38 were HIV positive without tuberculosis, and 30 were HIV negative without tuberculosis. Overall men and women in the study population had similar age regardless of HIV status except among HIV positive individuals without
178
tuberculosis; men were significantly older than women 34.7 ± 6.5 (SD) versus 29.7 ± 8.4
(SD), respectively.
Among HIV positive and HIV negative patients with tuberculosis, men and women had
comparable BMI whereas among HIV positive and HIV negative individuals without
tuberculosis, women had significantly higher BMI compared to men (Table 7:1 and 7:2).
Men had significantly higher fat-free mass and significantly lower fat mass compared to women regardless of tuberculosis and HIV status. Similarly, men had a higher magnitude of height-normalized FFMI and a lower height-normalized FMI compared to women regardless of tuberculosis and HIV status. Height-normalized FFMI and FMI eliminate differences in fat and fat-free mass associated with height (Baumgartner et al. 1998). Men with tuberculosis had significantly higher proportion of individuals with reduced FFMI compared to women regardless of HIV status (Table 7:1 and 7:2). For example among
HIV positive patients with tuberculosis, 7 out of 10 (70%) for men had reduced FFMI compared to 2 out of 21 (10%) for women (p<0.001). The clinical TB severity score was comparable between men and women regardless of HIV status. Regardless of gender on multiple comparisons, we found no differences in average BMI, fat and fat-free mass, and tuberculosis clinical severity score between HIV positive patients with tuberculosis and
HIV positive patients without tuberculosis, between HIV negative patients with tuberculosis and HIV negative individuals without TB, between HIV positive patients with tuberculosis and HIV negative patients with tuberculosis, and between HIV positive individuals without tuberculosis and HIV negative individuals without tuberculosis.
179
In general, there were no significant differences in average 24-hour dietary intake recall
between men and women for most nutrients except average energy, protein, total fat, and
folate among HIV negative individuals without tuberculosis; dietary fiber among HIV
positive individuals without tuberculosis; magnesium and zinc intake among HIV
negative patients with tuberculosis (Tables 7:3 and 7:4). Of note, however, men had a
high magnitude of average dietary intake for most nutrients, for example energy intake
was 2380 ± 703 for men versus 1598 ± 567 for women among HIV negative individuals
without tuberculosis, 2027 ± 1111 for men versus 1437 ± 920 for women among HIV
positive patients with tuberculosis (Table 7:3 and 7:4).
On average, BMI, fat mass, and height-normalized FMI were significantly lower among
tuberculosis patients with reduced fat-free mass, reduced BMI, and moderate/severe
TBscore compared to patients with normal fat-free mass, normal BMI, and mild TBscore.
However, the average severity TBscore was significantly higher among patients with reduced fat-free mass, reduced BMI, and moderate/severe TBscore (Table 7:5 and 7:6).
The average fat-free mass and height-normalized FFMI were similar between TB patients
with mild and patients with moderate/severe TBscore. However, patients with mild
TBscore had significantly higher proportion of individuals with reduced fat-free mass (16
out of 39 (41%)) compared to patients with moderate/severe TBscore (4 out of 24 (17%),
p <0.05) (Table 7:5 and 7:6).
180
Tuberculosis patients with moderate/severe TBscore had significantly lower average 24- hour dietary intake recall for average energy, protein, total fat, dietary fiber, calcium, vitamin A, and folate compared to patients with mild TBscore (Table 7:7 and 7:8). For example, energy intake among patients with moderate/severe TBscore was 1460 ± 722 compared to 2215 ± 940 among patients with mild TBscore. However, there were no significant differences in average 24-hour dietary intake recall between tuberculosis patients with reduced fat-free mass or reduced BMI and patients with normal fat-free mass or normal BMI (Table 7:7 and 7:8).
Discussion
In this cross-sectional study, we aimed to establish the relationship between dietary intake and body wasting; the independent effects of tuberculosis and HIV infection, and tuberculosis disease severity on dietary intake. In a study population of 131 HIV positive and HIV negative adults with/or without active tuberculosis from urban Uganda, we found that the 24-hour dietary intake recall varied by severity of tuberculosis disease, but not tuberculosis disease or HIV status. In the absence of tuberculosis, dietary intake varied by gender. The dietary intake differed by severity of clinical tuberculosis disease categories of mild and moderate/or severe disease. Men and women with tuberculosis had similar dietary intake regardless of HIV status. HIV negative women without tuberculosis had lower levels of dietary intake for energy, protein, and folate compared to HIV negative men without tuberculosis. Dietary intake did not differ by severity of nutritional status.
181
Findings of the present study suggest that in the face of tuberculosis disease, dietary
intake is affected by severity of disease but not HIV infection; and there is no association
with body wasting. While in the absence of tuberculosis, dietary intake is affected by
gender, and not HIV infection. Tuberculosis patients that had moderate/or severe clinical
disease had lower dietary intakes for energy, protein, total fat, carbohydrate, calcium,
vitamin A, and folate compared to patients with mild disease. Both men and women had comparable dietary intake among patients with tuberculosis regardless of HIV status whereas HIV negative women had reduced energy, protein, and folate intake among individuals without tuberculosis compared to men. Tuberculosis patients with fat-free
mass wasting or those with reduced BMI had comparable nutrient intakes with
counterparts that had normal fat-free mass or normal BMI. To our knowledge, this is the
first study to have evaluated the effect of tuberculosis disease severity on dietary intake
and whether there are differences in dietary intake between HIV positive and HIV
negative adults with tuberculosis. Our results are consistent with recent findings from
India in which the study (Swaminathan et al. 2008) found comparable nutrient dietary
intakes between HIV positive patients with tuberculosis and HIV positive individuals
without tuberculosis. This study however, had no comparable group of HIV negative
patients with tuberculosis to establish the synergist effect of tuberculosis and HIV
infection. The strengths of our study hinges on the full panel of HIV positive and HIV
negative adults with/or without tuberculosis. However, the study limited findings are
limited by the cross-sectional nature of the design that the associations are not causal
between differences between groups.
182
The present study demonstrated that dietary intake at the time of tuberculosis diagnosis was influenced by disease severity and in populations without tuberculosis was influenced by gender. Dietary intake differed by tuberculosis disease severity and not by tuberculosis disease or HIV status. While in populations without tuberculosis, nutrient intakes differed by gender particularly among HIV negative individuals. Women had reduced intakes for energy, protein, and folate. Despite high proportions of patients with body wasting as assessed by reduced fat-free mass (32% (20/63) with reduced fat-free mass) or BMI (56% (35/63) with reduced BMI), nutrient intakes were not associated with body wasting. Nutrient intakes were similar between individuals with wasting and those without. Two potential reasons can explain the findings in the present study. First, the disparity in reduced appetite between patients with mild clinical disease and those with moderate/or severe disease. Patients with moderate/or severe clinical disease present with marked low appetite that impends the nutrient intake other than barriers to food access.
Further, the comparability of appetite level among men and women with tuberculosis, and among patients with/or without body wasting explains the gender and the wasting status similarity in nutrient intake. One can postulate that tuberculosis disease affects the appetite level equally in men and in women regardless of body wasting status, thus, compromising nutrient intake at nearly the same rate regardless of food access barriers.
Thus, the gender differences in body composition during tuberculosis as revealed in the present and in previous reports (Mupere et al. 2010) could be explained by the differences in altered metabolism. Men generally, experience higher metabolic rate because of the larger quantities of fat-free mass compared to women (Arciero, Goran, and
183
Poehlman 1993) and during tuberculosis, this process is probably pronounced with associated wasting.
The gender differences in dietary intake among HIV negative individuals without tuberculosis could be explained by the cultural factors that may compromise intake among women. For example, unequal distribution of food within households (Carloni
1981; de Hartog A.P 1972), or men may have the opportunity to eat a wider variety or better quality foods outside the home, such as cafes or local restaurants (Holmboe-
Ottesen G and Wandel M 1991). The unequal distribution of food within households result from several factors such as women may be trained to show restraint in eating, to give the best foods to men, or to allow others in the family to eat first (Lado 1992;
O'Laughlin B 1974; Rosenberg E.M 1980; Dey J 1981).
To conclude, the study revealed that dietary intake at the time of diagnosis was influenced by tuberculosis disease severity, but not tuberculosis disease or HIV status and in the absence of tuberculosis was influenced by gender. Nutritional counseling and supplementation, early treatment and prevention of tuberculosis are needed to improve dietary intake in population of sub-Saharan Africa.
184
Acknowledgements
We thank all study staff members of the Case Western Reserve University and Makerere
University research collaboration at the Tuberculosis Research Unit in the United States and in Uganda for their assistance; the faculty of staff at Case Western Reserve
University Department of Epidemiology and Biostatistics for the guidance in analyzing the project; and the Fogarty International Center, for the continued support.
This study was supported in part by the AIDS International Training Research Program,
Fogarty International Center, Grant No. TW000011, based at Case Western Reserve
University, Department of Epidemiology and Biostatistics. This work was part of Ezekiel
Mupere’s PhD thesis at Case Western Reserve University.
185
Table 7:1 Select characteristics among HIV positive and HIV negative adults with tuberculosis
Characteristic HIV positive with TB HIV negative with TB
[mean, (SD)] (n=31) (n=32)
Men Women Men Women
(n=10) (n=21) (n=18) (14)
Age in yr 30.9 (4.6) 29.2 (5.9) 26.0 (7.3) 26.3 (4.6)
BMI kg/m2 18.4 (1.7) 18.6 (3.0) 18.2 (2.0) 20.3 (4.3)
BMI
<18.5 kg/m2 (%) 7 (70) 12 (57) 11 (61) 5 (36)
≥18.5 kg/m2 (%) 3 (30) 9 (43) 7 (39) 9 (64)
Fat-free mass in kg 49.1 (4.5) 38.8 (3.2)a 49.1 (4.7) 39.9 (5.6)a
FFMI in kg/m2 16.6 (1.3) 15.4 (0.9)b 16.6 (1.5) 16.1 (0.9)
FFMI
<16.7 for M, <14.6 for W (%) 7 (70) 2 (10)a 11 (61) 0 (0)a
≥16.7 for M, ≥14.6 for W (%) 3 (30) 19 (90) 7 (39) 14 (100)
Fat mass in kg 5.2 (1.8) 8.3 (5.2) 4.8 (2.3) 10.5 (6.6)b
186
FMI in kg/m2 1.8 (0.6) 3.3 (2.1) 1.6 (0.8) 4.6 (3.6)a
FMI
<1.8 for M, <3.9 for W (%) 5 (50) 14 (67) 11 (61) 8 (57)
≥1.8 for M, ≥3.9 for W (%) 5 (50) 7 (33) 7 (39) 6 (43)
Severity TBscore 6.5 (1.5) 5.6 (2.6) 6.8 (2.3) 5.9 (2.4)
Severity TBscore category
Mild ≤5 (%) 2 (20) 10 (48) 6 (33) 6 (43)
Moderate/severe >5 (%) 8 (80) 11 (52) 12 (67) 8 (57) ap-value <0.001, bp-value <0.05. BMI = body mass index, FFMI = fat-free mass index, FMI = fat mass index, SD = standard deviation, W = women, M = men.
187
Table 7:2 Select characteristics among HIV positive and HIV negative adults without tuberculosis
Characteristic HIV positive, no TB HIV negative, no TB
[mean, (SD)] (n=38) (n=30)
Men Women Men Women
(n=17) (21) (n=16) (14)
Age in yr 34.7 (6.5) 29.7 (8.4)b 22.4 (3.2) 24.3 (5.4)
BMI kg/m2 21.2 (2.2) 24.2 (4.6)b 21.6 (2.3) 23.7 (2.8)b
BMI
<18.5 kg/m2 (%) 2 (12) 5 (24) 1 (6) 1 (7)
≥18.5 kg/m2 (%) 15 (88) 16 (76) 15 (94) 13 (93)
Fat-free mass in kg 53.0 (4.8) 39.8 (3.0)a 51.9 (3.9) 42.5 (4.9)a
FFMI in kg/m2 18.3 (1.2) 16.6 (1.2)a 18.6 (1.4) 16.6 (1.1)a
FFMI
<16.7 for M, <14.6 for W (%) 2 (12) 0 (0) 1 (6) 1 (7)
≥16.7 for M, ≥14.6 for W (%) 15 (88) 21 (100) 15 (94) 13 (93)
Fat mass in kg 9.1 (3.2) 17.8 (8.2)b 7.9 (2.5) 18.2 (5.4)a
188
FMI in kg/m2 3.2 (1.2) 7.6 (3.7)a 2.9 (1.0) 7.1 (2.1)a
FMI
<1.8 for M, <3.9 for W (%) 3 (18) 5 (24) 1 (6) 1 (7)
≥1.8 for M, ≥3.9 for W (%) 14 (82) 16 (76) 15 (94) 13 (93) ap-value <0.001, bp-value <0.05. BMI = body mass index, FFMI = fat-free mass index, FMI = fat mass index, SD = standard deviation, W = women, M = men.
189
Table 7:3 Dietary intake of 24-hour recall among HIV positive and HIV negative adults with tuberculosis
Characteristic HIV positive with TB HIV negative with TB
[mean, (SD)]1 (n=31) (n=32)
Men (n=10) Women (n=21) Men (n=18) Women (n=14)
Energy (kcal) 2027 (1111) 1437 (920) 2094 (831) 1567 (511)
Protein (g) 60.6 (48.6) 49.5 (44.2) 56.0 (29.5) 42.8 (22.3)
Total fat (g) 59.5 (60.2) 32.4 (38.4) 56.5 (38.9) 39.8 (24.9)
Carbohydrate (g) 303 (158) 245 (141) 348 (139) 269 (82)
Dietary fiber (g) 42.7 (32.9) 30.2 (22.5) 41.4 (23.2) 39.8 (25.0)
Protein, % energy 11.9 (6.5) 13.2 (8.5) 10.7 (2.1) 10.7 (3.7)
Fat, % energy 23.4 (13.8) 15.9 (9.4) 22.5 (10.4) 21.1 (8.5)
CHO, % energy 64.7 (16.5) 70.7 (13.8) 66.9 (10.2) 68.3 (9.9)
Calcium (mg) 978 (1532) 558 (1029) 297 (319) 264 (200)
Magnesium (mg) 267 (267) 261 (250) 302 (160) 176 (134)b
Zinc (mg) 7.7 (5.9) 5.6 (4.5) 6.9 (3.2) 4.4 (2.1)b
Iron (mg) 11.9 (12.5) 11.5 (12.6) 11.8 (8.2) 9.0 (5.4)
190
Vitamin A (RE) 469 (568) 493 (108) 545 (835) 779 (745)
Ascorbic acid
(mg) 105 (64) 108 (140) 142 (159) 130 (80)
Vitamin D (μg) 1.1 (2.3) 1.0 (1.5) 0.6 (1.1) 0.6 (0.9)
Folate (μg) 300 (221) 315 (278) 346 (257) 375 (183) ap-value <0.001, bp-value <0.05. CHO = carbohydrate, 1Characteristic values are means ± standard deviation (SD). TB = tuberculosis
191
Table 7:4 Dietary intake of 24-hour recall among HIV positive and HIV negative adults without tuberculosis
Characteristic HIV positive, no TB HIV negative, no TB
[mean, (SD)]1 (n=38) (n=30)
Men (n=17) Women (21) Men (n=16) Women (14)
Energy (kcal) 1954 (699) 1651 (1105) 2380 (703) 1598 (567)a
Protein (g) 56.9 (34.5) 38.3 (24.8) 60.4 (21) 35.9 (16.3)a
Total fat (g) 47.2 (28.5) 34.0 (20.0) 60.7 (25.1) 38.2 (23.3)b
Carbohydrate (g) 333 (125) 301 (242) 416 (119) 274 (124)
Dietary fiber (g) 45.4 (27.8) 26.3 (16.8)b 40.1 (18.6) 23.3 (14.6)
Protein, % energy 11.2 (3.3) 9.7 (3.3) 9.8 (1.9) 8.9 (2.5)
Fat, % energy 20.8 (8.9) 19.9 (9.4) 21.4 (4.4) 20.5 (9.2)
CHO, % energy 68.0 (10.4) 70.6 (11.0) 68.7 (5.3) 67.7 (14.3)
Calcium (mg) 415 (414) 263 (404) 417 (454) 346 (406)
Magnesium (mg) 238 (204) 251 (336) 278 (181) 190 (150)
Zinc (mg) 7.1 (4.3) 5.7 (4.8) 7.2 (3.3) 3.8 (1.8)
Iron (mg) 11.1 (7.7) 11.5 (14.2) 12.4 (7.0) 7.0 (5.3)
192
Vitamin A (RE) 188 (238) 959 (1794) 936 (2716) 285 (314)
Ascorbic acid (mg) 129 (103) 142 (142) 144 (87) 86 (51)
Vitamin D (μg) 0.6 (1.2) 0.1 (0.4) 0.5 (0.8) 0.6 (0.8)
Folate (μg) 362 (174) 312 (190) 513 (218) 293 (148)b ap-value <0.001, CHO = carbohydrate, bp-value <0.05. 1Characteristic values are means ± standard deviation (SD). TB = tuberculosis
193
Table 7:5 Select characteristics among patients with/without wasting (n=63)
Fat-free mass BMI Characteristic Not wasted Wasted Not wasted Wasted [mean, (SD)] (n=43) (n=20) (n=28) (n=35)
Age in yr 28.1 (5.4) 27.4 (8.0) 28.0 (5.7) 27.8 (6.7)
BMI kg/m2 19.8 (3.2) 16.9 (1.2)a 21.3 (2.8) 16.9 (1.3)a
BMI
<18.5 kg/m2 (%) 16 (37) 19 (95)a - -
≥18.5 kg/m2 (%) 27 (63) 1 (5) - -
Fat-free mass in kg 42.2 (6.5) 46.2 (5.5)b 43.8 (7.3) 43.1 (5.7)
FFMI in kg/m2 16.3 (1.4) 15.6 (0.8)b 16.8 (1.4) 15.5 (0.8)a
FFMI
<16.7 for M, <14.6 for W (%) - - 1 (4) 19 (54)a
≥16.7 for M, ≥14.6 for W (%) - - 27 (96) 16 (46)
Fat mass in kg 8.8 (5.3) 4.1 (2.0)a 10.9 (5.3) 4.4 (2.2)a
FMI in kg/m2 3.6 (2.6) 1.4 (0.6)a 4.4 (2.8) 1.6 (0.9)a
FMI
194
<1.8 for M, <3.9 for W (%) 20 (47) 5 (25) 7 (25) 31 (89)a
≥1.8 for M, ≥3.9 for W (%) 23 (53) 15 (75) 21 (75) 4 (11)
Severity TBscore 5.6 (2.2) 8.0 (7.5)b 5.0 (2.1) 7.1 (2.1)a
ap-value <0.001, bp-value <0.05. FFMI = fat-free mass index, FMI = fat mass index, BMI = body mass index, W = women, M = men, and SD = standard deviation. Fat-free mass wasting = FFMI <16.7 kg/m2 for men and <14.6 kg/m2 for women, BMI wasting = <18.5 kg/m2 for men and women.
195
Table 7:6 Select characteristics among patients with/without severity of clinical tuberculosis
Severity TBscore Characteristic Mild ≤5 Moderate/severe >5 [mean, (SD)] (n=39) (n=24)
Age in yr 28.0 (6.8) 27.7 (6.0)
BMI kg/m2 20.2 (3.8) 18.0 (2.1)b
BMI
<18.5 kg/m2 (%) 24 (62) 11 (46)
≥18.5 kg/m2 (%) 15 (38) 13 (54)
Fat-free mass in kg 43.2 (7.0) 43.6 (6.1)
FFMI in kg/m2 16.3 (1.2) 15.9 (1.3)
FFMI
<16.7 for M, <14.6 for W (%) 16 (41) 4 (17)b
≥16.7 for M, ≥14.6 for W (%) 23 (59) 20 (83)
Fat mass in kg 9.6 (6.2) 5.9 (3.5)b
FMI in kg/m2 4.0 (3.2) 2.2 (1.4)b
196
FMI
<1.8 for M, <3.9 for W (%) 26 (67) 12 (50)
≥1.8 for M, ≥3.9 for W (%) 13 (33) 12 (50)
Severity TBscore 3.9 (1.3) 7.6 (1.6)a
ap-value <0.001, bp-value <0.05. FFMI = fat-free mass index, FMI = fat mass index, BMI = body mass index, W = women, M = men, and SD = standard deviation. Fat-free mass wasting = FFMI <16.7 kg/m2 for men and <14.6 kg/m2 for women, BMI wasting = <18.5 kg/m2 for men and women.
197
Table 7:7 Dietary intake of 24-hour recall among tuberculosis patients with/without body wasting (n=63)
Fat-free mass BMI Characteristic Not wasted Wasted Not wasted Wasted [mean, (SD)]1 (n=43) (n=20) (n=28) (n=35)
Energy (kcal) 1680 (790) 1893 (1070) 1727 (702) 1763 (1018)
Protein (g) 50.3 (33) 54.5 (44.3) 52.4 (36.8) 51.0 (37.3)
Total fat (g) 39.4 (32.3) 57.8 (54.2) 39.1 (22.8) 50.2 (50.9)
Carbohydrate (g) 288 (138) 290 (137) 296 (140) 283 (135)
Dietary fiber (g) 37.6 (24.4) 37.4 (27.0) 37.7 (25.1) 37.3 (25.4)
Protein, % energy 12.0 (6.5) 11.1 (4.5) 12.1 (6.6) 11.5 (5.4)
Fat, % energy 18.8 (9.6) 23.1 (12.0) 19.8 (9.0) 20.5 (11.7)
CHO, % energy 69.2 (12.2) 65.9 (12.8) 68.3 (11.6) 68.1 (13.3)
Calcium (mg) 448 (759) 565 (1125) 529 (895) 450 (885)
Magnesium (mg) 244 (193) 279 (242) 254 (161) 255 (242)
Zinc (mg) 5.7 (3.4) 6.8 (5.3) 5.5 (2.5) 6.5 (5.0)
Iron (mg) 11.2 (9.4) 10.9 (11.4) 10.3 (7.2) 11.7 (11.8)
198
Vitamin A (RE) 559 (608) 519 (807) 588 (708) 514 (650)
Vitamin D (μg) 0.9 (1.6) 0.7 (1.2) 0.8 (1.7) 0.8 (1.2)
Ascorbic acid (mg) 121 (112) 125 (150) 123 (126) 121 (125)
Folate (μg) 351 (227) 298 (270) 330 (225) 338 (256) ap-value <0.001, bp-value <0.05. 1Characteristic values are means ± standard deviation (SD). FFMI = fat-free mass index, FMI = fat mass index, BMI = body mass index, W = women, and M = men. CHO = carbohydrate, Fat-free mass wasting = FFMI <16.7 kg/m2 for men and <14.6 kg/m2 for women, BMI wasting = <18.5 kg/m2 for men and women.
199
Table 7:8 Dietary intake of 24-hour recall among tuberculosis patients with/without severe clinical TBscore (>5)
Severe TBscore Characteristic Mild ≤5 Moderate/severe >5 [mean, (SD)]1 (n=39) (n=24)
Energy (kcal) 2215 (940) 1460 (722)b
Protein (g) 70.5 (46.4) 40 (20.6)b
Total fat (g) 61.6 (50.3) 35.2 (30.6)b
Carbohydrate (g) 348 (130) 252 (128)b
Dietary fiber (g) 42.4 (27.4) 34.5 (23.4)
Protein, % energy 12.5 (7.6) 11.3 (4.6)
Fat, % energy 21.8 (10.8) 19.1 (10.3)
CHO, % energy 65.7 (13.4) 69.7 (11.8)
Calcium (mg) 772 (1330) 308 (344)b
Magnesium (mg) 312 (245) 220 (177)
Zinc (mg) 7.2 (5.0) 5.3 (3.3)
Iron (mg) 12.5 (10.9) 10.2 (9.5)
200
Vitamin A (RE) 751 (798) 421 (555)b
Vitamin D (μg) 0.7 (1.2) 0.9 (1.6)
Ascorbic acid (mg) 151 (131) 105 (119)
Folate (μg) 452 (283) 262 (179)b
ap-value <0.001, bp-value <0.05. 1Characteristic values are means ± standard deviation (SD). FFMI = fat-free mass index, FMI = fat mass index, CHO = carbohydrate, BMI = body mass index, W = women, and M = men. Fat-free mass wasting = FFMI <16.7 kg/m2 for men and <14.6 kg/m2 for women, BMI wasting = <18.5 kg/m2 for men and women.
201
CHAPTER 8
CORRELATES OF DIETARY INTAKE AMONG HIV POSITIVE AND HIV
NEGATIVE ADULTS WITH OR WITHOUT TUBERCULOSIS IN URBAN
KAMPALA, UGANDA
202
Abstract
Background Although understanding nutritional intake is important in the management of tuberculosis, its assessment in clinical and in research practice is often overlooked.
Objective We sought to determine correlates of energy and protein intake, and correlates
of inadequate dietary intake.
Methods In a cross-sectional study of 131 HIV positive and HIV negative adults with or
without tuberculosis who were enrolled from urban Uganda; 24-hour dietary intake recall
was assessed.
Results There was female gender interaction between having tuberculosis and reduced
appetite, and between having tuberculosis and being a current alcohol taker in the model
for energy intake. Women that had tuberculosis with reduced appetite or tuberculosis with history of taking alcohol had decreased energy intake. Also women who had history of alcohol intake had decreased protein intake. There was no compromise with energy and protein intake among men. Women were associated with inadequate iron intake.
Further, women with tuberculosis were associated with inadequate folate intake.
Individuals with tuberculosis residing in households of more than two people or those with no or low education were associated with inadequate vitamin A intake.
Conclusions The present study revealed that correlates of energy and protein intake
differ by gender. Women and individuals having tuberculosis who reside in overcrowded
households or who have no or low education are at vulnerable state of inadequate nutrient
intake. Further studies are needed to evaluate changes in nutrient intake and the impact
on survival.
203
Background
In sub-Saharan Africa, tuberculosis is the most common cause of death from a curable
infectious disease (Frieden et al. 2003). The incidence of clinical tuberculosis has
increased in sub-Saharan Africa due to the human immunodeficiency virus
(HIV/acquired immune-deficiency syndrome (AIDS) pandemic (Harries et al. 2001;
Lawn and Churchyard 2009), and despite adequate chemotherapy for tuberculosis,
mortality is still high among those co-infected with HIV (Haller et al. 1999).
Tuberculosis and HIV infections are both independently associated with body wasting and malnutrition. The wasting may be caused by a combination of decreased appetite, leading to a decrease in energy intake, interacting with increased losses and altered metabolism as part of the inflammatory and immune responses (Paton et al. 1999; Paton et al. 2003; Macallan et al. 1998). Moreover, wasting leads to impaired physical function
(Harries et al. 1988) and increased mortality in patients with tuberculosis (Mehta J.B et al. 1996; Rao et al. 1998; Zachariah et al. 2002; Mitnick et al. 2003). Although body wasting and malnutrition appear to separately play an important role in the clinical course of patients with tuberculosis and HIV and among those with co-infection, nutritional status and nutritional intake are often overlooked in clinical practice and in tuberculosis programs. Dietary intake studies are limited to characterize the nutritional intake, nutritional status and outcomes among patients with tuberculosis and HIV infection.
204
In the present cross-sectional study, we sought to determine correlates of energy and
protein intake, and correlates of inadequate dietary intake in a population of HIV positive
and HIV negative patients with or without tuberculosis enrolled from urban Kampala,
Uganda.
Methods
In a cross-sectional study, we enrolled 132 participants age 18 years or older residing in
Kampala district or 20 km from the study site if residence was outside Kampala in
Uganda. Data collection was conducted between November 2007 and March 2008, a period that coincides with harvesting and light rains in November and December and dry season in January and February. One participant was excluded from the analysis because of prior tuberculosis treatment. The study was conducted at the National Tuberculosis and Leprosy Program (NTLP) Clinic of the national tertiary teaching hospital, Mulago complex. Of the 131 participants who were included in the analysis, 31 were HIV positive and 32 were HIV negative with tuberculosis and were recruited at the Mulago
NTLP Clinic; 38 were HIV positive patients without tuberculosis and were recruited at the Infectious Disease Institute Clinic (IDI) located 500 meters from the Mulago NTLP
Clinic; and 30 were HIV negative individuals without tuberculosis and were recruited from the community where enrolled tuberculosis patients resided. The institutional review boards at Case Western Reserve University and Joint Clinical Research Center approved the study, with final approval by the Uganda National Council for Science and
Technology. All participants provided written informed consent to the study.
205
All subjects in the study were given appropriate pre- and post-test HIV counseling and
AIDS education. HIV-1 infection was diagnosed on the basis of a positive enzyme-linked immunosorbent assay for HIV-1 antibodies (Recombigen; Cambridge Biotech,
Cambridge, MA). At enrollment, basic demographic information and a medical history were collected, and a standardized physical examination was conducted by a medical officer. Active pulmonary TB was confirmed by sputum smear microscopy and culture.
Patients with active TB were treated with standard four-drug chemotherapy for tuberculosis per guidelines of the Ugandan Ministry of Health. Similarly, HIV positive patients eligible for antiretroviral therapy were started on treatment and cotrimoxazole prophylaxis at the IDI clinic.
The dietary intake assessment was made using a single 24-hour dietary recall with open ended questions. The reference period for the 24-hour recall was the day prior to the day of the interview. In all instances, the interview was held only with the interviewee, no one
else was present except for children. The questionnaire was pre-tested by administering it
to 8 individuals selected randomly from the neighboring community to the study site. The
assessment was conducted by four trained study nursing staffs and continuously
supervised by a nutritionist using local food photographs, portion-size images, and
volumetric vessels to increase the accuracy of the recall from the previous 24 hours. The
nutritive value of raw ingredients was computed using the East African food composition
table database whose database was imported into the NutriSurvey software
(http:www.nutrisurvey.de) to easy the computations. The database used was
predominantly for local Ugandan diet. When the East African food composition table was
206
found deficient in certain food items, the United States Department of Agriculture database and the African composition table were used.
To estimate the nutrient adequacy of the diet, we calculated the nutrient adequacy ratio
(NAR) (%) for 11 micronutrients, energy, protein, fat, carbohydrate, and dietary fiber.
The NAR for a given nutrient is the ratio of a participant’s intake to the daily recommended allowance for the participant’s sex. The Food and Agriculture
Organization/World Health Organization 2002 Human Vitamin and Mineral requirements
(FAO/WHO 2002) were used for vitamin A, vitamin B6, vitamin C, vitamin D, thiamin, riboflavin, folate, magnesium, calcium, iron, and zinc whereas energy, fat, protein, carbohydrate, and dietary fiber; the Panel on Macronutrients, Panel on the Definition of
Dietary Fiber, Submcommittee on Upper Reference Levels of Nutrients, Subcommittee on Interpretation and Uses of Dietary Reference Intakes Food and Nutrition Board
(Dietary Reference Intake for Energy, Carbohydrate, Fiber, Fat, Fatty Acids, Cholesterol,
Protein, and Amino Acids (Macronutrients). A report of the Panel on Macronutrients,
Subcommittees on Upper Reference Levels of Nutrients and Interpretation and Uses of
Dietary Reference Intakes, and the Standing Committee on the Scientific Evaluation of
Dietary Reference Intakes 2005) was used. The recommendation for adequate intake was used for vitamin D (Atkinson and Ward 2001). In the case of iron and zinc, the category for moderate bioavailability was used.
207
Analytic strategy
We performed bivariate, univariate, and multivariable analyses. Chi-square and Fisher’s
Exact tests were used to compare proportions. Fisher’s Exact test was used when tabular counts were less than 5. Mann-Whitney test was used to compare dietary intake variables due to lack of normality. To determine the independent correlates of absolute energy and protein intakes, multivariable linear regression analyses including all variables that were associated with a p < 0.50 in the unadjusted analyses (Dales and Ury 1978). The following variables were evaluated in unadjusted analyses: older age group >30 years, having tuberculosis and HIV infection, no or low level (primary) of education, being separated or divorced, being single, having a household number of more than 2 people, unemployment, having no personal income, current history of alcohol intake, and reduced appetite were evaluated.
We controlled for body size by including height in the regression model for energy intake because half of the participants had tuberculosis which could in itself lead to changes in intake. Furthermore, height was correlated with energy intake. The results of the multi-
regression analyses are presented along with the mean unadjusted intakes for each
subgroup variable considered. We adjusted for energy intake in all regression analyses of
protein intake to address the composition of the participant’s diets, independent of
absolute intake. The R square was used to establish variables that were important for the
reduced models. The residuals of the final models were found to be normally distributed
for energy and protein intake. We evaluated for two-way interactions between
208
tuberculosis or between HIV and all the important variables for both the energy and
protein intake models. Significant interactions were found between tuberculosis and
alcohol intake and between tuberculosis and reduce appetite in the energy intake model
for women.
Correlates of dietary inadequacy, i.e., intakes less than the recommended daily allowance were identified with preliminary unadjusted analyses by using linear models with a log link function and binomial distribution for prevalence response variable (Wacholder
1986). All variables that showed a p-value <0.50 in the unadjusted analyses for energy,
protein, vitamins A and D, folate, and zinc intakes were included in a multivariable full
model (Mickey and Greenland 1989). The Quasi-likelihood Information Criteria (QIC)
and Quasi-likelihood under the independence Criteria (QICu) goodness fit tests were
used in choosing the best models that fit the data. The analysis was performed using SAS
GENMOD procedure (SAS Institute 2003). We evaluated for two-way interactions
between tuberculosis, between HIV, or between gender and all the important variables for
models with energy, protein, vitamins A and D, folate, and zinc deficiency as main
outcome. Significant interactions were found between female gender and reduced
appetite for the model with protein deficiency, between tuberculosis and being more than
people per household for the model with vitamin A deficiency, and between tuberculosis
and gender for the model with folate deficiency as main outcomes. The prevalence ratios
and 95% confidence intervals (CIs) for each of the independent correlates of dietary
inadequacy on the basis of binomial regression models with log link function have been
209
reported in this paper (Mickey and Greenland 1989). All analyses were performed using
SAS version 9.2 (Cary software, North Carolina SAS Institute Inc 2004).
Results
The characteristics of the study population are shown in Table 8:1. Of the 131 participants who were included in the analysis, 53% were females, 47% were males, 53% were HIV positive, and 48% had tuberculosis. Women and men differed significantly by employment and income. Women had a higher proportion of individuals who were unemployed and without income compared to men. However, men had higher a proportion of individuals who were single compared to women.
The variables that were correlated with energy and protein intakes at a significance level of p < 0.50 in the unadjusted analysis are shown in Tables 8:2 and 8:3. For each of these variables, the mean energy and protein intakes are shown. None of the bivariate energy means was significantly different among both women and men. Also the bivariate protein means were comparable among women. There were significant interactions between having tuberculosis and reduced appetite (p=0.018), and between having tuberculosis and being a current alcohol taker in the model for energy intake among women (p=0.020)
(Table 8:2). Women that had tuberculosis and had reduced appetite or were current alcohol takers had decreased energy intake. Further, women who were current alcohol takers had decreased protein intake (Table 8:3). Energy and protein intakes were not
210
affected by any factor among men (Table 8:2 and 8:3). Of note, having HIV infection did influence energy and protein intakes regardless of gender.
The correlates of inadequate intakes of energy, protein, vitamin A and D, zinc, iron, and calcium are shown in Tables 8:4, 8:5, and 8:6. Intakes were considered inadequate if they were below the recommended daily allowance. Energy intake was inadequate in 73%
(96/131) of the participants. Inadequate intakes of protein, vitamin A and D, zinc, iron, folate, and calcium were found in 60%, 74%, 98%, 56%, 86%, 63%, and 95% of participants, respectively. Women were strongly associated with inadequate protein and iron intakes whereas men were associated vitamin A inadequate intake (Table 8:4 and
8:5). Participants with no or low level of education and participants that were not married
(singles), had inadequate vitamin A and iron intake, respectively. Furthermore, participants with reduced appetite were associated with inadequate vitamin A and folate intakes.
The multivariable analyses of the correlates of inadequate energy, protein, vitamins A and D, folate, and zinc intakes are shown in Table 8:6. The variables that correlated with inadequate intake for energy, protein, vitamins A and D, folate, and zinc at p < 0.50 in the unadjusted models were evaluated for their importance in the full model are presented in
Table 8:6 together with results of the reduced models. No factor was found to influence inadequate energy intake. Being a female was significantly associated with inadequate vitamin A and iron intake, and being a female with reduced appetite was associated with
211
inadequate protein intake. Having no or low level of education and being more than two
people per household in the face of tuberculosis were significantly associated with
inadequate vitamin A intake. Further, being a female with tuberculosis and having
reduced appetite were associated with inadequate folate intake. No factor influenced
inadequate vitamin D and zinc intakes.
Discussion
We sought to determine correlates of energy and protein intake, and correlates of inadequate dietary intake in this cross-sectional study of 131 HIV positive and HIV negative adults with or without tuberculosis who were enrolled from urban Uganda.
Correlates of energy and protein dietary intakes differed by gender. Protein intake among women was influenced by alcohol intake whereas energy intake among women with tuberculosis was influenced by reduced appetite or alcohol intake. Energy intake was not affected by any study variables. Gender, reduced appetite, level of education, and tuberculosis were important factors that were associated with inadequate intakes for vitamin A, iron, protein, and folate.
The caveats to the interpretation of results in the present study are not limited to the cross-sectional nature of the study but also use of a single 24-hour dietary recall in
nutritional assessment that we cannot comment on the seasonal effects or habitual
consumption. Findings in this study suggest the vulnerability of the female gender with
compromised nutrient intake and are at worse state of inadequacy in nutrient intake
212
compared to male gender. Also individuals in overcrowded households and with no or
low level of education in the face of tuberculosis are at worse state of nutrient
inadequacy. There was female gender interaction between having tuberculosis and
reduced appetite, and between having tuberculosis and being a current alcohol taker in
the model for energy intake. Women that had tuberculosis with reduced appetite or tuberculosis with history of taking alcohol had decreased energy intake. Also women who had history of alcohol intake had decreased protein intake. There was no compromise with energy and protein intake among men. Women were associated with inadequate iron intake. Further, women with tuberculosis were associated with inadequate folate intake. Individuals with tuberculosis residing in households of more than two people or those with no or low education were associated with inadequate vitamin A intake.
The gender differences in correlates of energy and protein dietary intake in this study population suggests the cultural factors that may compromise intake among women. The vulnerability of individuals with tuberculosis residing in households of more than two people or those with no or low education to inadequate vitamin A intake reflects on the poverty level of the community. For example, unequal distribution of food within households (Carloni 1981; de Hartog A.P 1972), or men may have the opportunity to eat a wider variety or better quality foods outside the home, such as cafes or local restaurants
(Holmboe-Ottesen G and Wandel M 1991). The unequal distribution of food within households result from several factors such as women may be trained to show restraint in
213
eating, to give the best foods to men, or to allow others in the family to eat first (Lado
1992; O'Laughlin B 1974; Rosenberg E.M 1980; Dey J 1981).
The present study revealed that correlates of energy and protein intake differ by gender.
Women and individuals having tuberculosis who reside in overcrowded households or who have no or low education are at vulnerable state of inadequate nutrient intake.
Further studies are needed to evaluate changes in nutritient intake and the impact on survival.
Acknowledgements
We thank all study staff members of the Case Western Reserve University and Makerere
University research collaboration at the Tuberculosis Research Unit in the United States and in Uganda for their assistance; the faculty of staff at Case Western Reserve
University Department of Epidemiology and Biostatistics for the guidance in analyzing the project; and the Fogarty International Center, for the continued support.
This study was supported in part by the AIDS International Training Research Program,
Fogarty International Center, Grant No. TW000011, based at Case Western Reserve
University, Department of Epidemiology and Biostatistics. This work was part of Ezekiel
Mupere’s PhD thesis at Case Western Reserve University.
214
Table 8:1 Characteristics of the study population (n=131)
Characteristics All subjects Women Men p-value (n=131) (n=70) (n=61)
n (%) n (%) n (%)
Age in years
≤30 94 (72) 52 (74) 42 (69) 0.491
>30 37 (28) 18 (26) 19 (31)
HIV status
Negative 62 (47) 28 (40) 34 (56) 0.072
Positive 69 (53) 42 (60) 27 (44)
Tuberculosis
No 68 (52) 35 (50) 33 (54) 0.640
Yes 63 (48) 35 (50) 28 (46)
Education
None/primary level 69 (53) 29 (41) 33 (54) 0.147
Secondary level 62 (47) 41 (59) 28 (46)
Tribe
215
Muganda 57 (44) 30 (43) 27 (44) 0.872
Others 74 (56) 40 (57) 34 (56)
Marital status
Married 56 (43) 29 (41) 27 (44) 0.744
Single 75 (57) 41 (59) 34 (56)
Household number
One to two 40 (31) 17 (24) 23 (38) 0.100
>2 people 91 (69) 53 (76) 38 (62)
Employed
No 54 (41) 36 (51) 18 (30) 0.011
Yes 77 (59) 34 (49) 43 (70)
Income
Not at all 47 (36) 35 (50) 12 (80) <0.001
Yes 84 (64) 35 (50) 49 (20)
Takes alcohol
No 98 (75) 51 (73) 47 (77) 0.581
Yes 33 (25) 19 (27) 14 (23)
216
Appetite
No 42 (32) 27 (61) 15 (75) 0.087
Yes 89 (68) 43 (39) 46 (25)
217
Table 8:2 Correlates of dietary energy intake among women and men
Characteristics Women Multivariate Men Multivariate model (n=70) (n=61) model
Mean Estimate Mean Estimate
(SD) (SE) (SD) (SE)
Age in years
≤30 1437 (683) 2184 (849)
>30 1815 (1153) 425.0 (235.0) 1976 (729) -
HIV status
Negative 1583 (530) 2229 (775)
Positive 1544 (1010) - 1981 (854) -297.4 (234.8)
Tuberculosis
No 1630 (917) 2161 (723)
Yes 1489 (776) 698.3 (341.5)b 2070 (920) -
Education
Secondary level 1546 (950) 2121 (841)
None/primary level 1579 (689) -46.4 (209.9) 2117 (803) 110.7 (224.5)
218
Marital status
Married 1731 (783) - 2133 (863)
Single 1439 (878) 2108 (786) 126.1 (231.6)
Household number
One to two 13228 (974) 2080 (682)
>2 1634 (797) 355.1 (239.3) 2143 (892) -
Income
Yes 1637 (796) 2087 (912)
Not at all 1482 (899) 118.0 (204.5) 2127 (798) -
Takes alcohol
No 1491 (776) 2060 (813)
Yes 1744 (1013) 682.9 (298.1)b 2319 (813) 238.7 (250.5)
Appetite
Normal 1463 (989) 1862 (770)
Reduced 1621 (749) 830.3 (498.8) 2203 (818) -370.1 (250.0)
TB*reduced - -425.2 (584.0)b - - appetite
219
TB*alcohol intake -1161.0 - - - (484.5)b
R2 0.28 0.10 ap-value <0.001, bp-value <0.05; bivariate p-values. Energy intakes were adjusted for height. Multivariate models show only those variables for which p < 0.50 in the unadjusted analyses.
220
Table 8:3 Correlates of dietary protein intake among women and men
Women Multivaiate Multivariate Characteristics Men (n=61) (n=70) model model
Mean Estimate Mean Estimate
(SE) (SD) (SD) (SE)
Age in years
≤30 36.6 (23.9) 59.0 (27.3)
>30 52.1 (43.2) 3.0 (6.5) 56.1 (41.9) -
Tuberculosis
No 37.3 (21.6) 58.6 (28.4)
Yes 46.8 (36.7) 10.5 (5.4) 57.6 (36.6) -
Education
No or primary level 45.5 (25.6) 60.0 (37.4)
Secondary level 39.6 (33.3 - 56.0 (25.3) -5.2 (5.0)
Tribe
Muganda 37.8 (22.7) 60.9 (35.5)
Others 45.3 (34.8) -8.2 (5.5) 56.0 (29.7) -
221
Marital status
Married 45.2 (25.8) 61.0 (39.8)
Single 39.9 (33.2) - 55.9 (25.0) 3.4 (5.3)
Household number
One to two 42.7 (47.1) 54.3 (18.7)
>2 41.9 (23.1) - 60.4 (38.2) 4.6 (5.6)
Takes alcohol
No 44.7 (34.0) 58.3 (35.2)
Yes 35.1 (15.0) -13.9 (6.2)b 57.5 (20.0) -
Appetite
No 43.7 (34.1) 61.5 (31.2)b
Yes 39.4 (23.4) - 47.7 (33.9) -4.1 (6.0)
R2 0.50 0.67
ap-value <0.001, appetite (p=0.036) bp-value <0.05; bivariate p-values. Protein intakes were adjusted for energy intakes. Multivariate models show only those variables for which p < 0.50 in the unadjusted analyses.
222
Table 8:4 Proportions of adult individuals with inadequate dietary intakes of energy, protein, and micronutrients in relation to socio-demographic, HIV, and tuberculosis variables in Kampala, Uganda
Vitamins Characteristics Energy Protein A D
Gender
Female 50/70 (71) 48/70 (69)b 46/70 (66)b 69/70 (99)
Male 46/61 (75) 30/61 (49) 51/61 (84) 60/61 (98)
Age in years
≤30 69/94 (73) 55/94 (59) 69/94 (73) 93/94 (99)
>30 27/37 (73) 23/37 (62) 28/37 (76) 36/37 (97)
HIV status
Negative 43/62 (62) 36/62 (58) 47/62 (76) 62/62 (100)
Positive 53/69 (77) 42/69 (61) 50/69 (72) 67/69 (97)
Tuberculosis
No 50/68 (74) 38/68 (56) 54/68 (79) 68/68 (100)
Yes 46/63 (73) 40/63 (63) 43/63 (68) 61/63 (97)
223
Education
Secondary level 43/62 (69) 35/62 (56) 41/62 (66)b 61/62 (98)
None/primary 43/69 (62) 56/69 (81) 53/69 (77) 68/69 (99) level
Tribe
Muganda 41/57 (72) 35/57 (61) 45/57 (79) 56/57 (98)
Others 55/74 (74) 43/74 (58) 52/74 (70) 73/74 (99)
Marital status
Married 39/56 (70) 33/56 (59) 36/56 (64)b 55/56 (98)
Single 57/75 (76) 45/75 (60) 61/75 (81) 74/75 (99)
Household number
One to two 32/40 (80) 55/91 (60) 60/91 (66) 90/91 (99)
>2 64/91 (70) 23/40 (58) 37/40 (93) 39/40 (98)
Employed
Yes 59/77 (77) 47/77 (61) 58/77 (75) 75/77 (97)
No 37/54 (69) 31/54 (77) 39/54 (72) 54/54 (100)
Income
224
Yes 64/84 (76) 51/84 (61) 63/84 (75) 82/84 (98)
Not at all 32/47 (68) 27/47 (57) 34/47 (72) 47/47 (100)
Takes alcohol
No 73/98 (74) 57/98 (58) 71/98 (72) 96/98 (98)
Yes 23/33 (70) 21/33 (64) 26/33 (79) 33/33 (100)
Appetite
Normal 62/89 (70) 48/89 (54) 71/89 (80)b 87/89 (98)
Reduced 30/42 (71) 26/42 (62) 42/42 34/42 (81) (1000 ap-value <0.001, bp-value <0.05; bivariate p-values.
225
Table 8:5 Proportions of adult individuals with inadequate dietary intakes of energy, protein, and micronutrients in relation to socio-demographic, HIV, and tuberculosis variables in Kampala, Uganda
Minerals Characteristics Zinc Iron Folate Calcium
Gender
Female 41/70 (59) 66/70 (94)b 46/70 (66) 67/70 (96)
Male 33/61 (54) 47/61 (77) 36/61 (59) 57/61 (93)
Age in years
≤30 51/94 (54) 81/94 (86) 56/94 (60) 90/94 (96)
>30 23/37 (62) 32/37 (86) 26/37 (70) 34/37 (92)
HIV status
Negative 36/62 (58) 53/62 (85) 34/62 (55) 60/62 (97)
Positive 38/69 (55) 60/69 (87) 48/69 (70) 64/69 (93)
Tuberculosis
No 38/68 (56) 58/68 (85) 39/68 (57) 65/68 (97)
Yes 36/63 (57) 55/63 (87) 43/63 (68) 59/63 (59)
226
Education
Secondary level 37/62 (60) 51/62 (82) 40/62 (65) 59/62 (95)
None/primary level 37/69 (54) 62/69 (90) 42/69 (61) 65/69 (94)
Tribe
Muganda 32/57 (56) 48/57 (84) 38/57 (67) 53/57 (93)
Others 42/74 (57) 65/74 (88) 44/74 (59) 71/74 (96)
Marital status
Married 29/56 (52) 50/56 (89) 35/56 (63) 53/56 (95)
Single 45/75 (60) 63/75 (84) 47/75 (63) 71/75 (95)
Household number
One to two 52/91 (57) 81/91 (89) 58/91 (64) 85/91 (93)
>2 22/40 (55) 32/40 (80) 24/40 (60) 39/40 (98)
Employed
Yes 44/77 (57) 65/77 (84) 49/77 (49) 73/77 (95)
No 30/54 (56) 48/54 (88) 33/54 (61) 51/54 (94)
Income
Yes 45/84 (54) 69/84 (82) 56/84 (67) 78/84 (93)
227
Not at all 29/47 (62) 44/47 (94) 26/47 (55) 46/47 (98)
Takes alcohol
No 56/98 (57) 87/98 (89) 63/98 (64) 94/98 (96)
Yes 18/33 (55) 26/33 (79) 19/33 (58) 30/33 (91)
Appetite
Normal 48/89 85/89 (96) 49/89 (55) 76/89 (85) (54)b
Reduced 25/42 (60) 37/42 (88) 34/42 (81) 39/42 (93) ap-value <0.001, bp-value <0.05; bivariate p-values.
228
Table 8:6 Correlates of inadequate dietary intakes of key nutrients from multivariate models among adult individuals in Kampala, Uganda
Characteristics Full model1 Reduced model2
Prevalence ratio p- Prevalence ratio p-
(95% CI) value (95% CI) value
Inadequate energy intake
HIV positive 1.07 (0.85, 1.34) 0.58 1.07 (0.85, 1.34) 0.58
No education/primary level 1.08 (0.87, 1.34) 0.62 1.08 (0.87, 1.34) 0.62
Single 0.91 (0.73, 1.14) 0.42 0.91 (0.73, 1.14) 0.42
>2 per household 0.91 (0.73, 1.14) 0.43 0.91 (0.73, 1.14) 0.43
No income 0.93 (0.73, 1.17) 0.52 0.93 (0.73, 1.17) 0.52
Reduced appetite 1.16 (0.94, 1.43) 0.16 1.16 (0.94, 1.43) 0.16
QIC 900.8 900.8
Inadequate protein intake
Female 1.34 (0.99, 1.81) 0.06 1.62 (1.08, 2.43) 0.020
Tuberculosis 0.97 (0.68, 1.40) 0.88 0.97 (0.68, 1.40) 0.88
No education/primary level 1.05 (0.79, 1.40) 0.73 1.06 (0.80, 1.40) 0.69
229
Single 1.29 (0.90, 1.85) 0.17 0.98 (0.74, 1.30) 0.87
Reduced appetite 0.96 (0.72, 1.28) 0.79 1.81 (1.81, 3.02) 0.023
Female*reduced appetite - - 0.59 (0.34, 1.03) 0.063
QIC 564.1 554.6
Inadequate Vitamin A
Female 0.80 (0.66, 0.98) 0.029 0.75 (0.61, 0.92) 0.01
Tuberculosis 0.95 (0.76, 1.18) 0.62 1.35 (1.04, 1.74) 0.02
No education/primary level 1.28 (1.05, 1.57) 0.015 1.28 (1.05, 1.56) 0.01
Other than Muganda 1.19 (0.98, 1.44) 0.08 1.16 (0.96, 1.40) 0.14
Single 0.84 (0.68, 1.04) 0.11 0.83 (0.68, 1.02) 0.08
>2 per household 0.77 (0.65, 0.92) 0.004 1.03 (0.79, 1.34) 0.84
Takes alcohol 1.00 (0.82, 1.24) 0.96 1.06 (0.87, 1.29) 0.57
Reduced appetite 0.83 (0.64, 1.09) 0.17 0.85 (0.66, 1.10) 0.23
TB*>2 per household - - 0.56 (0.39, 0.82) 0.003
QIC 929.8 919.0
Inadequate folate
Female 1.12 (0.85, 1.48) 0.42 1.57 (1.02, 2.42) 0.04
230
Age >30 years 1.13 (0.87, 1.47) 0.35 1.21 (0.91, 1.62) 0.19
HIV positive 1.17 (0.87, 1.58) 0.30 1.09 (0.83, 1.43) 0.52
Tuberculosis 0.99 (0.70, 1.39) 0.94 0.82 (0.64, 1.06) 0.13
No education/primary level 0.84 (0.65, 1.09) 0.19 0.99 (0.77, 1.27) 0.94
Other than Muganda 1.02 (0.79, 1.34) 0.83 0.73 (0.50, 1.09) 0.12
Single 0.94 (0.72, 1.22) 0.65 0.97 (0.75, 1.26) 0.81
No income 0.82 (0.60, 1.12) 0.21 0.83 (0.61, 1.13) 0.23
Reduced appetite 1.45 (1.03, 2.03) 0.03 1.48 (1.07, 2.04) 0.02
TB*female - - 1.92 (1.13, 3.28) 0.02
QIC 603.1 600.8
Iron
Female 1.20 (1.03, 1.40) 0.02 1.20 (1.03, 1.40) 0.02
Single 1.05 (0.91, 1.22) 0.49 1.05 (0.91, 1.22) 0.49
>2 people per household 1.03 (0.85, 1.25) 0.74 1.03 (0.85, 1.25) 0.74
Unemployment 0.91 (0.73, 1.13) 0.40 0.91 (0.73, 1.13) 0.40
No income 1.16 (0.93, 1.44) 0.20 1.16 (0.93, 1.44) 0.20
Takes alcohol 0.89 (0.75, 1.06) 0.19 0.89 (0.75, 1.06) 0.19
231
QIC 1800.4 1800.4
Vitamin D
HIV positive 0.95 (0.90, 1.01) 0.13 0.95 (0.90, 1.01) 0.13
Tuberculosis 0.92 (0.83, 1.03) 0.13 0.92 (0.83, 1.03) 0.13
No education/primary level 1.01 (0.97, 1.06) 0.57 1.01 (0.97, 1.06) 0.57
Unemployment 1.00 (0.98, 1.03) 0.81 1.00 (0.98, 1.03) 0.81
No income 1.03 (0.99, 1.08) 0.65 1.03 (0.99, 1.08) 0.65
Takes alcohol 1.02 (0.99, 1.04) 0.22 1.02 (0.99, 1.04) 0.22
Reduced appetite 1.09 (0.98, 1.22) 0.13 1.09 (0.98, 1.22) 0.13
QIC 16188.0 16188.0
Inadequate zinc intake
Age >30 years 1.24 (0.89, 1.72) 0.21 1.24 (0.89, 1.72) 0.21
No education/primary level 0.88 (0.65, 1.20) 0.43 0.88 (0.65, 1.20) 0.43
Single 0.86 (0.63, 1.17) 0.33 0.86 (0.63, 1.17) 0.33
No income 1.16 (0.86, 1.58) 0.33 1.16 (0.86, 1.58) 0.33
QIC 520.7 - 520.7 -
1The adjusted analyses first controlled for all variables that showed an association with inadequate intake of the outcome variable at p < 0.50 in unadjusted models. 2Shows best model.
232
CHAPTER 9
IMPACT OF BODY WASTING ON SURVIVAL AMONG ADULT PATIENTS
WITH PULMONARY TUBERCULOSIS IN URBAN KAMPALA, UGANDA
233
Abstract
Background Body mass index (BMI) may over estimate body wasting and mortality due to wasting because it is insensitive to body fatness at low BMI and above normal muscle development. We assessed the impact of HIV and body wasting on survival in tuberculosis patients using precise measures of nutritional status, the height-normalized fat-free mass (FFMI) and fat mass (FMI) indices.
Methods In a retrospective cohort study of adult patients, 310 patients with baseline wasting and 437 without wasting as measured by BMI (kg/m2); 103 with baseline
wasting and 208 without wasting as measured by fat-free mass index (kg/m2); and 401
HIV positive and 346 HIV negative patients were followed for survival.
Results During the follow-up period, 19% of 310 patients with baseline wasting by BMI
died compared to 11% of 437 without wasting, a crude risk ratio of 1.74 (95% confidence
interval (CI): 1.22, 2.48). Of 103 with baseline wasting by fat-free mass index, 16% died,
compared to 7% without wasting, crude risk ratio of 2.31 (95% CI: 1.10, 3.92). In
stratified survival analysis, survival proportion was significantly lower among men with
reduced BMI compared to men with normal BMI; and lower among women with reduced
fat-free mass index compared to women with normal fat-free mass index. In multivariable Cox regression model using anthropometric data, the relative hazard of death when patient had reduced BMI was 1.85 (95% CI: 1.25, 2.73). In a nested model, the relative hazard for death was 1.70 (95% CI: 1.03, 2.81) for men with reduced BMI and 1.83 (95% CI: 0.96, 3.50) for women with reduced BMI. In a model using fat-free
234
mass index data, the relative hazard of death when patients had reduced fat-free mass index was 1.88 (0.96, 3.65). In a nested model, the relative hazard of death was 6.83
(95% CI: 2.14, 21.74) for women with reduced fat-free mass index compared to women with normal fat-free mass and 0.80 (95% CI: 0.35, 1.84) for men with reduced fat-free mass index. In Kaplan-Meier analysis, men had significantly lower survival compared to women (p=0.016) Cox regression analysis HIV positive men had 1.62 (95% confidence interval (CI): 1.05, 2.52) hazard of death compared to HIV positive women. HIV negative men had 0.57 (95% CI: 0.10, 3.31) hazard of death compared to HIV negative women.
Conclusion Findings show that body wasting is associated with reduced survival that differed by gender. BMI is a better predictor of death among men whereas fat-free mass index among women. There was gender differences in survival with poor outcomes among HIV positive men.
235
Background
Tuberculosis is among the foremost infectious cause of mortality worldwide. Globally,
recent reports showed 9.2 million new cases and 1.7 million tuberculosis-associated
deaths to have occurred in 2006 (Vitoria et al. 2009). In HIV-infected individuals, tuberculosis is the leading cause of death accounting for up to 11% of AIDS-related mortality worldwide (Corbett et al. 2003; Holmes et al. 2003). In sub-Saharan Africa
where there is a high burden of both TB and human immunodeficiency virus (HIV)
infection, case fatality rates for tuberculosis in HIV-infected patients are extremely high
(up to 40%) (Corbett et al. 2003; Mugusi et al. 2009). Tuberculosis is frequently found at
autopsy of acquired immunodeficiency syndrome (AIDS) patients (Lucas et al. 1993),
particularly cachectic patients suggesting that tuberculosis exacerbates the wasting
process of HIV-infected people in Africa.
Body wasting is regarded as a cardinal feature of tuberculosis. A significant proportion of
African tuberculosis patients have a marked degree of wasting by the time they present
for registration and treatment (Kennedy et al. 1996; Harries et al. 1988; Zachariah et al.
2002). Wasting associated with tuberculosis is likely caused by a combination of
decreased appetite, leading to a decrease in energy intake, interacting with increased
losses and altered metabolism resulting from the inflammatory and immune responses
(Paton et al. 1999; Paton et al. 2003; Macallan 1999). Wasting is associated with
impaired physical function (Harries et al. 1988), longer hospitalization days and
236
increased mortality in patients with tuberculosis (Rao et al. 1998; Zachariah et al. 2002;
Mitnick et al. 2003). It has been reported that in patients with both TB and HIV co-
infection, the wasting process is exacerbated (Macallan 1999; Lucas et al. 1994). In
contrast, findings from several cross-sectional studies appear to show no large differences
in body composition between HIV-infected adults with tuberculosis and HIV-negative
adults with tuberculosis at presentation (Shah et al. 2001; Niyongabo et al. 1999;
Niyongabo et al. 1994; Mupere et al. 2010) suggesting that TB is the dominant factor
inducing wasting. This has been shown in several reports (Mupere et al. 2010; Paton and
Ng 2006). However, gender differences in body composition at presentation among TB
patients have been reported (Kennedy et al. 1996; Mupere et al. 2010).
Although the association of malnutrition with tuberculosis, and its impact on co-infection
with tuberculosis and HIV has been described (Macallan 1999; van Lettow, Fawzi, and
Semba 2003) in clinical settings and in several studies that have evaluated mortality in
tuberculosis (Zachariah et al. 2002; Mugusi et al. 2009), the extent of wasting has been
assessed using body mass index (BMI). However, BMI is insensitive to body fatness,
particularly at low BMI, as well as with above-normal muscle development (Kyle,
Genton, and Pichard 2002; Kyle, Piccoli, and Pichard 2003). Thus, previous estimates of
mortality between malnourished and individuals with normal nutrition might have been
overestimated. Furthermore, the confounding effects of gender and HIV infection on
survival among tuberculosis patients using precise estimates of body composition have
not yet been described. Fat and fat-free mass body composition measurements have been shown to permit a more precise evaluation of nutritional status (VanItallie et al. 1990;
237
Kyle, Piccoli, and Pichard 2003). Bioelectrical impedance analysis (BIA) has been
recommended as the practical and precise method for clinical assessment of fat and fat-
free mass (Kyle et al. 2004; Kyle, Genton, and Pichard 2002). In this study, we evaluated
the impact of body wasting using fat and fat-free mass and BMI as measures of body wasting; and the impact of HIV infection on survival among patients with pulmonary tuberculosis. Our results show for the first time gender differences of observed survival using fat-free mass.
Methodology
Study Design
We conducted a retrospective cohort study that consisted of 753 adult pulmonary tuberculosis patients having confirmed HIV status and defined baseline body wasting or not using the completed five year Household Contact (HHC) study, the completed phase
II prednisolone double blind randomized placebo controlled clinical trial, and the ongoing
Kawempe Community Health (KCH) study. Of the 753 patients, 314 were enrolled into
the HHC, 344 into the KCH, and 95 into the placebo arm of the prednisolone clinical
trial. The HHC and KCH studies were observational epidemiologic studies; organized
and conducted by the Makerere University and Case Western Reserve University
tuberculosis research collaboration (Uganda-CWRU) that has been ongoing for the last
20 years in Uganda. The HHC was the initial household contact study from 1995 to 1999 that described the epidemiology of tuberculosis in urban Kampala, Uganda (Guwatudde et al. 2003). The subsequent KCH started in 2002 and is still ongoing (Stein et al. 2005).
238
The KCH was developed specifically to focus on the determinants of host factors
associated with primary infection, re-infection, reactivation, and progression of clinical disease and to identify and track individual strains of mycobacterial tuberculosis through
Ugandan households and local community. The phase II clinical trial was conducted
between 1995 to 2000 to determine whether immunoadjuvant prednisolone therapy in
HIV-infected patients with tuberculosis who have CD4(+) T cell counts >/=200 cells/ mu
L is safe and effective at increasing CD4(+) T cell counts.
The institutional review boards at Case Western Reserve University in the United States
and Joint Clinical Research Center in Uganda reviewed the protocol and final approval
was obtained from the Uganda National Council for Science and Technology. All
patients in the HHC and KCH had written informed consent to be enrolled in the study.
All participants in both HHC and KCH were given appropriate pre- and post-test HIV counseling and AIDS education. HIV-1 infection was diagnosed on the basis of a positive enzyme-linked immunosorbent assay for HIV-1 antibodies (Recombigen; Cambridge
Biotech, Cambridge, MA). HIV-seropositive participants who were newly identified with
HIV were not on antiretroviral therapy at the time of measurement; no patients with pre- existing HIV infection at the time of household evaluation were on antiretroviral therapy because patients were enrolled before the advent of antiretroviral therapy in Uganda.
During follow-up, patients who became eligible were started on antiretroviral therapy.
239
At enrollment, basic demographic information and a medical history were collected, and
a standardized physical examination was conducted by a medical officer. Active
tuberculosis was confirmed by sputum smear microscopy and culture. Patients with
active tuberculosis were treated with standard four-drug chemotherapy for tuberculosis per guidelines of the Ugandan Ministry of Health. Adults with a previous history of treated pulmonary tuberculosis were excluded in the study. Of the 753 participants who were enrolled in the three studies, 6 were excluded because of being below 18 years of age, leaving 747 participants in total available for anthropometric analysis.
The datasets for the three studies were first tested for differences in baseline characteristics before combining for analysis. The datasets were different regarding extent of tuberculosis disease on chest x-ray because the phase II prednisolone trial enrolled only HIV-associated tuberculosis patients with CD4 cell count >200 cells/l compared to HHC and KCH studies (Appendix, Table 13:1). One of our interests was to establish the confounding effect of HIV; therefore, we combined all the three datasets for analysis. The BIA data (specifically, fat-free mass and fat mass) were collected during the KCH study only. Of the 344 participants who were enrolled in KCH, 33 were excluded due to lack of BIA measurements (n=29) data and being below 18 years of age
(n=4), leaving 311 participants in total available for BIA analysis. However, there were no differences in baseline age, gender, weight, height, BMI, smoking status, hemoglobin, chest x-ray disease extent, and history of weight loss between participants who were included and those who were excluded.
240
Measurements
In all the three studies, socio-demographic and clinical information was obtained through
standardized interviews and physical examination performed by trained medical officers.
Venous blood was collected for HIV-1 enzyme immunoassay testing and complete blood
and differential counts. HIV infection was documented by enzyme-linked immunosorbent
assays. All participants had posterior-anterior chest X-rays taken at baseline.
Expectorated sputum specimens were collected, concentrated, and stained for acid fast bacilli (AFB) with Ziehl-Neelsen stain at the Wandegeya national reference laboratory in
Uganda. AFB smears were reviewed by trained technicians who graded the smears by the number of acid-fast organisms seen on the light microscopy according to criteria established by the WHO (International Union Against Tuberculosis and Lung Disease
1986). Specimens were cultured for mycobacteria tuberculosis on Lowenstein-Jensen
medium slants, incubated at 370C in air and examined weekly until positive or for 8
weeks.
Nutritional status was assessed using anthropometric measurements such as height and
weight and BIA Detroit, MI, RJL Systems. Body weight was determined to the nearest
0.1 kg using a SECA adult balance, and standing height was determined to the nearest 2 mm. Body-mass index (BMI) was computed using the relationship of weight in kilograms divided by height in meters squared (kg/m2). All BIA measurements were
performed by one trained observer using the same equipment and recommended standard
conditions with regard to body position, previous exercise, dietary intake, skin
241
temperature, and voiding of the bladder were taken into consideration in taking BIA
measurements (Kyle et al. 2004). All BIA measurements during the KCH study were
performed on the day patients were confirmed to have tuberculosis disease.
The BIA is a simple, easy, safe, non-invasive technique, that has been recommended for nutritional studies in the clinical setting (Kyle et al. 2004; Kyle, Piccoli, and Pichard
2003) and is a convenient method to determine the lean or fat-free mass and fat body compartments (Kyle, Piccoli, and Pichard 2003; Kyle et al. 2004). Single-frequency BIA was performed at 50 kHz and 800 mA with standard tetrapolar lead placement (Jackson et al. 1988). Before performing measurements on each participant, the BIA instrument was calibrated using the manufacturer’s recalibration device. The resistance and reactance were based on measures of a series circuit (Kotler et al. 1996). BIA measurements were performed in triplicate for each subject. Fat-free mass was calculated from BIA measurements using equations that were previously cross-validated in a sample of patients (white, black and Hispanic) with and without HIV infection (Kotler et al.
1996) and have been applied elsewhere in African studies (Villamor et al. 2006; Shah et al. 2001; Van Lettow et al. 2004). Fat mass was calculated as body weight minus fat-free
mass.
Exposure variable
Baseline body wasting was the main exposure. We used BMI and height-normalized indices (adjusted for height2) of body composition that partition BMI into fat-free mass
242
index (FFMI) and fat mass index (FMI) (Schutz, Kyle, and Pichard 2002; VanItallie et al.
1990; Kyle, Piccoli, and Pichard 2003) to establish the body wasting status of
participants. The FFMI and FMI have the advantages of compensating for differences in
height and age (Kyle, Genton, and Pichard 2002). Also, the use of the FFMI and FMI
eliminates some of the differences between population groups. We defined body wasting
as patients having the low fat-free mass index (FFMI) and the low body fat mass index
(FMI) corresponding to WHO BMI categories for malnutrition as previously reported
(Table 9:1) (Kyle, Piccoli, and Pichard 2003). The FFMI <16.7 (kg/m2) for men and
<14.6 (kg/m2) for women and the FMI <1.8 (kg/m2) for men and <3.9 (kg/m2) for women
corresponds to a BMI of <18.5 kg/m2, the WHO cutoff for malnutrition (WHO Tech Rep
1995) among adults.
Table 9:1 Definitions of low and normal fat and fat-free mass index values for corresponding body mass index in adults
Characteristic Low Normal
Body mass index (BMI)a
Women and men in kg/m2 < 18.5 ≥ 18.5
Fat-free mass index (FFMI)b
Women in kg/m2 < 16.7 ≥ 16.7
Men in kg/m2 < 14.6 ≥ 14.6
243
Fat mass index (FMI)b
Women in kg/m2 < 1.8 ≥ 1.8
Men in kg/m2 < 3.9 ≥ 3.9
aWorld Health Organization categories, sex independent (WHO Tech Rep 1995). bKyle et al (Kyle, Piccoli, and Pichard 2003; Kyle et al. 2003).
Study outcome variable
Observed survival was the main study outcome. We defined observed survival as the
period between enrollment in the study and death or censoring. Participants for the three
studies were censored at the last clinic visit when they were known to be alive or at the
end of the study observation. During the conduct of the three studies at the Makerere
University-Case Western Reserve University research collaboration, mortality was assessed through a standard interview of family members or review of hospital records.
When a participant failed to keep a scheduled visit, the health visitor made a visit to the participant’s home to determine the vital status of the study participant. The family members also provided the date of death and prominent symptoms at the time of death.
Statistical analysis
Baseline characteristics for both participants with and participants without baseline wasting were compared using the χ2 test or Fisher’s exact test (where tabular counts were
less than 5) for binary data and student’s t-test for continuous variables or Wilcoxon-
244
Mann Whitney test for variables not normally distributed. Concordance in estimating wasting between the BMI cut-off of <18.5 kg/m2 and the FFMI cut-off of <16.7 (kg/m2) for men and <14.6 (kg/m2) for women, and the concordance in estimating wasting between BMI cut-off level and the FMI cut-off of <1.8 (kg/m2) for men and <3.9 (kg/m2) for women was assessed using kappa, к. Where к = 0 represents no agreement, 0.- 0.2 represents slight agreement, 0.2 – 0.4 represents fair agreement, 0.4 – 0.6 represents moderate agreement, 0.6 – 0.8 represents substantial agreement, and 0.8 – 1.0 represents almost perfect agreement, beyond chance (McGinn et al. 2004).
Mortality rates were estimated and stratified according to sex, young and old age group,
HIV status, anemia, smoking status, alcohol intake status, history of weight loss, and presence or absence of moderate/ or far advanced disease extent on chest X-ray. The incident rate of death was estimated using person-years method for the study population.
The overall distributions of survival for participants presenting with or without body wasting were estimated using the Kaplan-Meier method and compared using the log-rank test (Kaplan E.I and Meier P 1958). Overall the observed range of the data, the probabilities of death did not reach 50% so the median time from diagnosis of tuberculosis with body wasting as assessed by reduced BMI and FFMI to death could not be estimated.
A series of Cox proportional hazards models (Hosmer D.W.Jr and Lemeshow S 1999) were fit after testing for the proportional hazards assumptions for each variable. The
245
proportional hazards assumption was tested by graphical methods and goodness of fit
Schoenfeld residuals (Appendix, Table 13:2). Observed survival was the depended
variable in each model. The independent variables tested included age, sex, HIV status,
presence or absence of anemia, smoking and alcohol intake status, presence or absence of
history of weight loss, and presence or absence of moderate/or far advanced tuberculosis
disease extent on chest X-ray. Variables that were associated with survival in a univariate
analysis with p < 0.30 or with biological plausibility were evaluated in a series of multivariable models.
All two-way interactions between baseline body wasting status as measured by reduced
BMI or FFMI and the main effects were compared; the only significant interaction was with sex in the models involving FFMI variable for body wasting status. Furthermore, in our previous study (Mupere et al. 2010) we found that there are gender differences in body composition among patients with pulmonary tuberculosis. Thus, we developed separate stratified Cox models for participants according to sex status. Using a chunk wise test, interactions with sex, HIV, and age were found to be important in the model for reduced BMI whereas sex, age, and hemoglobin were important in the model for reduced
FFMI. Hence separate stratified Cox regression models were performed according to HIV status and according to young (≤30 years) or older (>30 years) age groups to understand the independent effect of HIV and age on hazard of death. All variables fulfilled the proportional hazard assumption within the gender, HIV status, and age group strata.
Models were compared using -2 log likelihood tests. All analyses were performed using
SAS version 9.1.3 (Cary software, North Carolina SAS Institute Inc. 2000 – 2004).
246
Results
Descriptive statistics
Of the 747 patients who were included in the anthropometric data analysis, 395 (53%) were men and 352 (47%) were women; 401 (54%) were HIV positive, 345 (46%) were
HIV negative, and 1 had unknown HIV status; and the mean age was 30.8 ± 8.9 SD years. Women had significantly higher BMI (20.3 ± 3.3 versus 18.6 ± 2.1) compared to men (Table 9:2). Of the 311 patients included in the BIA analysis, 164 (53%) were men and 147 (47%) were women; 138 (45%) were HIV positive, 172 (55%) were HIV negative, and 1 had unknown HIV status; and the mean age was 29.9 ± 9.2 SD years.
Women had significantly lower FFMI (15.8 ± 1.1 versus 16.6 ± 1.4) and significantly higher FMI (4.4 ± 2.8 versus 2.0 ± 0.8) compared to men (Table 9:2). Among men, FFMI had a strong positive correlation with BMI whereas among women, FMI had a strong positive correlation with BMI. These gender differences in body composition and correlation with BMI reflect on the high fat body content among women and the high fat- free mass among men as previously reported (Mupere et al. 2010). Furthermore, the strengths of correlation between BMI and FMI suggest that BMI is a measure of body fatness.
Among women, there was fair concordance [к = 0.34, (95% CI: 0.18, 0.50)] between
FFMI (women <14.6 kg/m2) and BMI (<18.5 kg/m2) cut-off for assessing body wasting with 34% of the women having both reduced FFMI and BMI whereas among men there was moderate concordance [к = 0.57, (95% CI: 0.45, 0.69)] with 80% of participants
247
having both reduced FFMI (for men <16.7 kg/m2)) and BMI (<18.5 kg/m2) (Table 9:3).
Fat mass index (FMI <3.9 kg/m2) and BMI cut-off for body wasting among women had substantial concordance [к = 0.61, (95% CI: 0.49, 0.73)] with 100% of participants having both reduced FFMI and BMI whereas among men, the concordance was moderate
[к = 0.52, (95% CI: 0.39, 0.65)] with 64% of participants having both reduced FMI (<1.8 kg/m2) and BMI (Table 9:3). Of the 747 patients who were included in the
anthropometric data analysis, 310 (42%) had body wasting at presentation and 437 (58%)
no wasting based on BMI cut-off of < 18.5 kg/m2 for both men and women. In the study
population of 311 participants that had BIA measurements, 103 (33%) had body wasting
at presentation and 208 (67%) no wasting using FFMI cut-off of <16.7 kg/m2 for men
and <14.6 kg/m2 for women. Whereas using FMI cut-off of <3.9 kg/m2 for women and
for <1.8 kg/m2 for men, 135 (43%) had baseline wasting and 176 (57%) had no wasting.
Baseline characteristics
Men had a higher frequency of reduced BMI and FFMI compared to women at presentation (Table 9:4). For example men had a proportion of 63% (194/310) for reduced BMI compared to 37% (116/310) among women. Men had a proportion of 80%
(82/103) of reduced FFMI compared to 20% (21/103) in women. Patients who presented with anemia (hemoglobin ≤10 mg/dl), history of prior smoking and current alcohol intake had a lower frequency of reduced BMI and FFMI. Patients who presented with moderate/far advanced disease on chest x-ray and history of weight loss had a higher frequency of reduced BMI and FFMI (Table 9:4). Of note, there were no differences in
248
body wasting at presentation between HIV positive and HIV negative tuberculosis patients regardless of whether BMI, FFMI, and FMI cut-off was used in assessing wasting (Appendix, Table 13:3). This suggests the lack of HIV impact on tuberculosis body wasting as previously reported (Mupere et al. 2010). HIV positive patients had a significantly higher frequency of older individuals (64% (200/312) versus 46%
(201/434)), anemic individuals (74% (89/121) versus 51% (173/342)), current alcohol intake (62% (174/279) versus 49% (226/466), and lower frequency of history of weight loss (51% (295/578) versus 64% (105/165)) than HIV negative patients at presentation.
Survival analysis
During the mean follow-up period of 31 ± 23 (SD) months for the anthropometric data, there were a total of 105 deaths. Of these 105 deaths, 76 occurred within the first 6 months during tuberculosis treatment, 18 occurred in the first year after completion of tuberculosis treatment at 6 months, and only 11 occurred more than one year after treatment. The overall annual mortality rate was 55.1 per 100 years of observation. The mortality rates were 393.6 per 100 person months of observation in the first 6 months,
326.9 per 100 person years of observation in the first year following treatment, and 6.0 per 100 person years of observation more than one year following treatment. Most deaths
(n=99) occurred among tuberculosis patients with HIV infection. In addition to HIV status, wasting as measured by BMI was also common (n=58). Of the 310 patients who presented with reduced BMI ≤ 18.5 kg/m2 19% died compared to 11% of the 437 patients
249
with normal BMI >18.5 kg/m2, leading to a crude risk ratio of 1.74 (95%: CI, 1.22 –
2.48) (Table 9:5).
Among patients with reduced BMI, 32% of the 167 that had HIV infection died
compared to 20% who died of the 234 that had HIV infection with normal BMI at
presentation, leading to a relative risk of 1.61 (95% CI: 1.50, 2.27) (Table 9:5). Among
HIV negative tuberculosis patients; however, the relative risk of death associated with
body wasting at presentation was 7.06 (95% CI: 0.83, 59.81) compared with patients
without wasting (Table 9:5). The relative risk of death associated with body mass wasting
at presentation among tuberculosis patients increased significantly after stratifying by
younger age group ≤30 years, hemoglobin >10 mg/dl, non-current alcohol intake status, and normal/or mild disease extent on chest x-ray (Table 9:5).
During the mean follow-up period of 26 ± 16 (SD) months for the BIA data, there were a total of 30 deaths. Of these 30 deaths, 15 occurred within the first 6 months during tuberculosis treatment, 13 occurred in the first year after completion of tuberculosis treatment at 6 months, and only 2 occurred more than one year after treatment. The overall annual mortality rate was 45.1 per 100 years of observation. The mortality rates were 141.4 per 100 person months of observation in the first 6 months, 283.3 per 100 person years of observation in the first year following treatment, and 3.3 per 100 person years of observation more than one year following treatment. Most deaths (n=29) occurred among tuberculosis patients with HIV infection.
250
In addition to HIV status, wasting as measured by FFMI was also common (n=16). Of the
103 patients who presented with reduced FFMI 16% died compared to 7% of the 208
patients with normal FFMI, leading to a crude risk ratio of 2.31 (95%: CI, 1.17 – 4.54)
(Table 9:6). Among patients with reduced FFMI, 32% of the 47 that had HIV infection
died compared to 15% who died of the 91 that had HIV infection with normal FFMI at
presentation, leading to a relative risk of 2.07 (95%: CI, 1.10 – 3.92) (Table 6). However,
co-infected patients with reduced FFMI had a higher mortality of 32% (15/47) compared
to 15% (14/91) among co-infected patients with normal FFMI, leading to a relative risk
of 2.07 (95% CI: 1.10, 3.92) (Table 9:6). The relative risk of death associated with body wasting at presentation among tuberculosis patients increased after stratifying by female sex, younger age group <30 years, anemia (hemoglobin ≤10 mg/dl), non -current smoking, and normal/or minimal disease extent on chest X-ray (Table 9:6). The relative risk of death associated with fat-free mass wasting at presentation among tuberculosis patients increased significantly after stratifying by female sex, normal/or mild extent of x-ray disease, anemia (hemoglobin ≤10 mg/dl), history of past smoking, and history of weight loss. The risk of death among women with fat-free mass wasting was dramatic with an estimate of 6.00 (95% CI: 2.14, 16.86) (Table 9:6).
Kaplan-Meier analysis
When we performed a Kaplan-Meier analysis using anthropometric data, the overall unadjusted survival for patients who presented with body wasting (BMI <18.5 kg/m2) had significantly lower survival compared to patients who presented with normal nutrition
251
(BMI ≥ 18.5 kg/m2) (p = 0.001, log-rank test). When the Kaplan-Meier analysis was stratified according to gender, survival proportion was significantly lower among men with reduced BMI <18.5 kg/m2 at presentation compared to men with normal BMI ≥ 18.5
kg/m2 at presentation (p = 0.033, log-rank test; Figure 9:1). For women with reduced
BMI at presentation, the survival proportion was lower than women with normal BMI,
but this difference was not statistically significant (p = 0.119, log-rank test; Appendix
Figure 13:1).
When we performed a Kaplan-Meier analysis using the BIA data, the overall unadjusted
survival for patients who presented with fat-free mass body wasting (FFMI <16.7 kg/m2
for men and <14.6 kg/m2 for women) had significantly lower survival compared to
patients who presented with normal nutrition (FFMI ≥16.7 kg/m2 for men and ≥14.6
kg/m2 for women) (p = 0.016, log-rank test). When the Kaplan-Meier analysis was stratified according to sex, survival proportion was significantly lower among women with reduced fat-free mass (FFMI <18.5 kg/m2) at presentation compared to women with
normal fat-free mass (FFMI ≥14.6 kg/m2) at presentation (p = 0.0004, log-rank test;
Figure 9:2). For men with reduced FFMI at presentation, the survival proportion was not
different from men who had normal FFMI (p = 0.647, log-rank test; Appendix Figure
13:2).
When we performed a Kaplan-Meier analysis according to sex status, men had
significantly lower unadjusted survival compared to women (p=0.035, log-rank test;
252
Appendix Figure 13:3). When the Kaplan-Meier analysis was stratified according to HIV status, survival proportion was significantly lower among HIV positive men compared to
HIV positive women (p = 0.016, log-rank test; Figure 9:3). For HIV negative men, the survival proportion was not different from HIV negative women (p = 0.937, log-rank test;
Appendix Figure 13:4).
Cox Proportional Hazard Regression
In the univariate Cox proportional hazards model, the unadjusted relative hazard for death in patients with reduced BMI <18.5 kg/m2 at presentation compared to patients who presented with normal BMI ≥18.5 kg/m2 was 1.80 (95% CI, 1.23, 2.64). Similarly, the unadjusted relative hazard for death in patients with reduced FFMI (<16.7 kg/m2 for men and <14.6 kg/m2 for women) at presentation compared to patients who presented with normal FFMI (≥16.7 kg/m2 for men and ≥14.6 kg/m2 for women) was 2.34 (95% CI,
1.14, 4.80; Table 9:7). Furthermore, men, older age group >30 years, HIV positive status, anemia (hemoglobin ≤10 mg/dl), and history of weight loss at presentation were significantly associated with increased relative hazards of death (Table 9:7).
In multivariable Cox proportional hazards model for anthropometric data, after adjusting for age, HIV, current smoking status, history of weight loss, and chest X-ray disease extent, the relative hazard for death when patients had reduced BMI <18.5 kg/m2 at presentation was 1.85 (95% CI, 1.25, 2.73; Table 9:8 and 9:9). Basing on our prior findings and the descriptive statistics presented in this paper that body composition differ
253
by gender (Mupere et al. 2010), we fitted separate nested Cox regression models with the same covariates according to gender (Table 9:9). Men with reduced BMI <18.5 kg/m2 at presentation had a significantly 1.70 (95% CI: 1.03, 2.81) times greater relative hazard of death than men with normal BMI ≥18.5 kg/m2. A 1.83 (95% CI: 0.96, 3.50) relative
hazard of death was found among women presenting with reduced BMI compared to
women presenting with normal BMI; however, this was not significant. The differences
in the relative hazard of death suggest interaction across the gender stratum. Older
patients >30 years of age among men presenting with reduced BMI also had significantly
greater relative hazard of death compared to younger men. In addition, HIV positive and patients with history of weight loss presenting with reduced BMI had significantly greater relative hard of death compared to HIV negative and patients without history of weight loss regardless of gender, respectively (Table 9:9).
In a multivariable model with reduced BMI as main predictor and after adjusting for age, sex, prior smoking status, history of weight loss, and extent of disease on chest x-ray,
patients with reduced BMI had a 1.75 (95% CI: 1.18, 2.60) times greater hazard of death
compared to individuals with normal BMI. Among HIV positive patients in stratified Cox
regression, individuals with reduced BMI had a significant 1.63 (95% CI: 1.09, 2.44)
times greater hazard of death compared to individuals with normal BMI whereas among
HIV negative patients, patients with reduced BMI had a 6.95 (95% CI: 0.78, 61.89) times
greater hazard of death compared to patients with normal BMI (Table 9:10 and 9:11).
This suggests interaction according to HIV status when patients present with/or without
reduced BMI. There were also potential interactions according to gender, age group, and
254
prior history of smoking. HIV positive men, HIV positive older patients, and HIV positive patients with prior history of smoking had significant hazards of death compared to HIV positive women, HIV positive young patients, and HIV positive patients with no prior history of smoking (Table 9:11). The converse was true that HIV negative men,
HIV negative older patients, and HIV negative patients with prior history of smoking had similar hazards of death compared to HIV negative women, HIV negative young patients, and HIV negative patients with no prior history of smoking. There were no significant interactions between the reduced BMI variable for wasting and all the variables adjusted for in the model (Appendix, Table 13:4). However, interactions with male sex, positive
HIV status, and older age group were found to be important in the reduced BMI multivariable model (Appendix, Table 13:5).
In a multivariable model with reduced BMI as main predictor and after adjusting for HIV positive status, male sex, prior smoking status, history of weight loss, and extent of disease on chest x-ray, patients with reduced BMI had a 1.78 (95% CI: 1.20, 2.64) times greater hazard of death compared to patients with normal BMI. Among young patients in stratified Cox regression, individuals with reduced BMI had a significant 2.85 (95% CI:
1.41, 5.77) times greater hazard of death compared to individuals with normal BMI whereas among old patients, individuals with reduced BMI had a 1.39 (95% CI: 0.86,
2.24) times greater hazard of death compared to patients with normal BMI (Appendix,
Tables 13:6, 13:7, 13:8, and 13:9). This also suggests interaction according to age group stratum of young (≤30 years) and old (>30 years) when patients present with/or without reduced BMI.
255
In the multivariable Cox proportional hazards model using reduced FFMI as main predictor after adjusting for age, HIV, and anemia (hemoglobin ≤10 mg/dl) the relative hazard for death when patients had reduced FFMI (<16.7 kg/m2 for men and <14.6 kg/m2 for women) at presentation was 1.88 (95% CI, 0.96, 3.65; Table 9:12 and 9:13). When sex, HIV, age, and hemoglobin were included in this model, there was a statistically significant interaction between patients presenting with reduced FFMI and gender
[hazard ratio (HR), 0.122 (95% CI, 0.030, 0.56)]. Furthermore, basing on our prior findings that body composition differ by gender (Mupere et al. 2010), we fitted separate stratified Cox regression models with the same covariates according to gender (Table
9:13). Women with reduced FFMI <14.6 kg/m2 at presentation had nearly 7 fold greater relative hazard of death than women with normal FFMI ≥14.6 kg/m2. While men presenting with reduced FFMI <16.7 kg/m2 had a protective hazard of death of 0.80 (95%
CI, 0.35, 1.84) compared to men presenting with normal FFMI ≥16.7 kg/m2, although this was not significant. Older patients >30 years of age among women presenting with reduced FFMI also had significantly greater relative hazard of death compared to young women. In addition, HIV positive patients presenting with reduced FFMI had significantly greater relative hazard of death compared to HIV negative patients regardless of gender (Table 9:13).
Interactions of reduced FFMI variable with HIV, age, and hemoglobin in the model were not significant; however, they were found to be important in the model (Appendix,
Tables 13:12 and 13:13). In a multivariable model with reduced FFMI as main predictor and after adjusting for age, sex, and hemoglobin, individuals with reduced BMI had a
256
1.50 (95% CI: 0.73, 3.08) times greater hazard of death compared to individuals with
normal FFMI. Among HIV positive patients in stratified Cox regression, individuals with
reduced FFMI had a non-significant 1.50 (95% CI: 0.72, 3.12) times greater hazard of
death compared to individuals with normal FFMI whereas among HIV negative patients,
the hazard of death for presenting with reduced FFMI could not be estimated because of
very events (Appendix, Tables 13:14 and 13:15). This suggests that HIV is not a
confounder when patients present with/or without reduced BMI.
In a multivariable model with reduced FFMI as main predictor and after adjusting for
HIV positive status, male sex, and hemoglobin, patients with reduced BMI had a non-
significant 1.55 (95% CI: 0.76, 3.15) times greater hazard of death compared to patients
with normal FFMI. Among young patients in stratified Cox regression, individuals with
reduced FFMI had a non-significant 1.89 (95% CI: 0.55, 6.52) times greater hazard of
death compared to individuals with normal FFMI. Similarly, among old patients,
individuals with reduced FFMI had a non-significant 1.52 (95% CI: 0.63, 3.69) times greater hazard of death compared to patients with normal FFMI (Appendix, Tables 13:13
– 13:16). This also suggests that age is not a confounder when patients present with/or without reduced FFMI.
There was no significant relative hazard of death between fat mass body wasting and any of the main effect variables in both univariate, multivariable, and stratified Cox models.
257
Discussion
To our knowledge, this is the first study to report the effect of body composition on
survival among tuberculosis patients and the involved gender differences. In this
retrospective cohort study of adult patients with pulmonary tuberculosis in urban Uganda,
body wasting as measured by reduced fat-free mass and body mass indexes was
associated with reduced survival, but the effect varied according to gender of the patient.
Among men with reduced BMI <18.5 kg/m2 at diagnosis, the effect of body mass wasting
was substantial whereas in women with reduced BMI, the effect was minimal. Among
women with reduced fat-free mass index (FFMI <14.6 kg/m2) at diagnosis, the effect of
fat-free mass wasting was dramatic with a 7 fold risk of death whereas among men presenting with reduced FFMI, the effect on observed survival was protective. The current results indicate that reduced body mass is a predictor of survival among men whereas reduced fat-free mass is a predictor of survival among women and the effect of
wasting is greatest early in the first year following diagnosis of tuberculosis. Fat mass
wasting appears not to be predictor of survival regardless of gender.
This study found gender differences in the measures of body wasting and malnutrition
and the corresponding effect on observed survival. Compared to FFMI, BMI was found
to be a better measure of body wasting associated with the observed survival in men
whereas FFMI was better measure in women. This appears to suggest that BMI may
underestimate mortality in women whereas FFMI may underestimate mortality in men. A
possible explanation stems from the fact that BMI is composed of both fat and fat-free
258
mass each of which influence BMI in the same direction concerning mortality as
previously hypothesized (Allison et al. 1997). However, fat and fat-free mass may influence mortality in opposite directions; that is, fat-free mass may be protective and fat mass deleterious. Though BMI is monotonically related to adiposity (Garrow and
Webster 1985), it also correlates positively with the amount of fat-free mass an individual has. In our study, BMI had a significant positive correlation of 0.85 in men and 0.58 in women. Thus, the inverse gender differences in fat and fat-free mass explains the differences in correlation. The high hazard of death reported in this study among men and among women may be an indicator of excessively low amount of fat mass in men and low amount of fat-free mass in women regardless of BMI. However, body fat mass wasting was not associated with risk of death regardless of among women and men.
The findings of the present appear to suggest that the effects of body wasting and malnutrition on tuberculosis are not apparent when there is no comorbidity, yet tuberculosis is associated with malnutrition (Macallan 1999; Schwenk A and Macallan
D.C 2000). In the face of comorbidities such as HIV and anemia, the effects of body wasting and malnutrition become detectable because the force of mortality from comorbidity overwhelms the existing malnutrition. Results of the present study and similar to previous studies in sub-Saharan Africa (Shah et al. 2001; Zachariah et al.
2002), show that more than 30% of tuberculosis patients present with body wasting and malnutrition regardless of gender or HIV status, and whether BMI or BIA parameters were used in assessing nutritional status. Furthermore in the present study, there were no differences in body wasting between HIV positive and HIV negative patients at the time
259
of diagnosis suggesting that tuberculosis may be the driving factor in inducing the
wasting process and the role of HIV is minimal. This finding is consistent with previous
studies (Mupere et al. 2010; Paton and Ng 2006). However overtime in the present study, results revealed heterogeneity in the effects of HIV on survival in the presence of body
wasting as measured by reduced BMI. Patients that presented with reduced BMI (<18.5
kg/m2) and were HIV positive had a significant 1.63 (95% CI: 1.09, 2.44) hazard of death
compared to patients that presented with normal BMI and were HIV positive. Yet there
was minimal effect on observed survival among patients that presented with reduced BMI
and were HIV negative. Malnutrition in itself is a cause of immunodepression
(Hernandez-Pando, Orozco, and Aguilar 2009); thus, tuberculosis might worsen the
course of HIV-associated immunodepression. These interrelated effects possibly explain
the facts that both tuberculosis and malnutrition are associated with reduced survival
among HIV-infected patients (Nunn et al. 1992; Suttmann et al. 1995).
In this study, more than 75% of deaths occurred early in the first year of follow-up. There are several potential reasons for the early deaths of tuberculosis patients in the present study. First, late presentation with severe and extensive tuberculosis; more than 75% of the participants had moderate/or far advanced disease on chest x-ray. Second, most
(>90%) deaths were HIV-related; yet HIV-tuberculosis co-infection is associated with extra burden of nutritional alterations that may lead to poor outcomes. The extra burden on nutritional status may include increased energy expenditure, nutrient malabsorption, reduced intake, micronutrient malnutrition, and increased production of inflammatory cytokines with lipolytic and proteolytic activity (van Lettow, Fawzi, and Semba 2003;
260
Niyongabo et al. 1994; Melchior et al. 1993). Third, about 25% of the study population
presented with anemia (hemoglobin ≤10 mg/dl); yet anemia is a life-threatening HIV-
related complication that may be associated with poor outcome. Finally, wasting status at
diagnosis; we have shown that body wasting regardless of whether assessed by BMI or
fat-free mass index was associated with an increased hazard of death which could be early or late.
Findings in the present study show gender differences in survival and the survival was modified by HIV infection. HIV positive men had poor survival compared to HIV positive women; however, HIV negative men and HIV negative women had similar survival. Our findings suggest that in the face of co-infection, survival is poor among
HIV positive men while in the absence of HIV, survival is similar by gender. Our findings are similar to previous studies of mortality in HIV-infected patients in Africa in which the rate of mortality was higher in men than in women (Lucas et al. 1993; Sani et al. 2006). Our study however, had a large sample size and a heterogeneous study population that included ambulatory and inpatients compared to previous studies (Sani et al. 2006; Lucas et al. 1993) enrolled only in-patients with probable advanced disease. The poor survival among HIV positive men could be due to delay in presentation for care among men. Prior studies have explained the late presentation among men to be due to a false sense of security concerning relative risk HIV infection (Luginaah, Yiridoe, and
Taabazuing 2005). Further studies are needed to evaluate the socioeconomic and biologic factors that predispose HIV positive men to poor survival.
261
The caveat to the interpretation of findings in this study include 1) the method (BIA) used in measurement of body composition is not of reference standard like the dual-energy x- ray absorptiometry, 2) the BIA prediction method used has not yet been validated in the local population. As a result, findings of body composition may be biased because of variations in hydration across ethnic groups (Kyle et al. 2004). However, the equations that were used in this study were previously cross validated in individuals of different race (white, black, and Hispanic) among men and women, who were both healthy controls and HIV-infected patients (Kotler et al. 1996). Moreover, the equations have been used widely in other studies from Africa with meaningful findings (Shah et al. 2001;
Van Lettow et al. 2004; Villamor et al. 2006; Mupere et al. 2010). We also took care to take measurements at rest, with proper placement of leads, in participants who had not exercised or taken alcohol, in participants with voided bladder and ambient temperature.
However, measurements were in patients with underlying illness that may cause shifts in body water compartments, thereby affecting measurements of fat mass. Our findings also limited by the lack of dietary intake assessment to give further insight in the interpretation of gender differences in body composition and observed survival.
Despite the limitations of the present study, the strengths of this study are that a large number of patients were studied, long duration of follow-up period to observe survival, and assessing of nutritional status using both BMI and the precise measures of fat and fat- free mass.
262
The findings in this study indicate that body wasting exerts greatest effect on observed survival among tuberculosis patients with body wasting co-infected with HIV, and that
BMI is a better predictor of death among men whereas FFMI is a better predictor of death among women. These observations provide evidence for the need to provide nutritional interventions among HIV co-infected patients with wasting and evaluation of nutritional status should be involve BMI, fat and fat-free mass.
Acknowledgements
We thank all study staff members of the Case Western Reserve University and Makerere
University research collaboration at the Tuberculosis Research Unit in the United States and in Uganda for their assistance; the faculty of staff at Case Western Reserve
University Department of Epidemiology and Biostatistics for the guidance in analyzing the project; and the Fogarty International Center, for the continued support.
This study was supported in part by the AIDS International Training Research Program,
Fogarty International Center, Grant No. TW000011, based at Case Western Reserve
University, Department of Epidemiology and Biostatistics, and Tuberculosis Research
Unit (grant N01-AI95383 and HHSN266200700022C/ N01-AI70022 from the NIAID).
This work was part of Ezekiel Mupere’s PhD thesis at Case Western Reserve University.
263
Table 9:2 Mean body mass, fat and fat-free mass indexes; spearman correlations
between body mass index and fat or fat-free mass indexes among adult women and men in urban Uganda
Mean (SD)a Correlations with BMIb
Characteristic Women Men Women Men
(n=147) (n=164) (n=147) (n=164)
1BMI, kg/m2 20.0 (3.3) 18.6 (2.1) 1 1
FFMI, kg/m2 15.8 (1.1) 16.6 (1.4) 0.58 0.85
FMI, kg/m2 4.4 (2.8) 2.0 (0.8) 0.92 0.68
BMI = body mass index, FFMI = fat-free mass index, FMI = fat mass index. 1Three hundred fifty two women and 401 men were involved in estimation of mean BMI; aall p- values were <0.001 for mean differences between women and men; ball correlations between FFMI or FMI and BMI among women and men were significant with p-values <0.001.
264
Table 9:3 Concordance between low body mass index and low fat-free or fat mass indexes corresponding to body mass index in assessing wasting among adults in urban Uganda
Women
Characteristic Kappa, к Normal BMI aReduced BMI (95% CI)
FFMI index kg/m2b
Reduced, n (%) 5/100 (5) 16/47 (34) 0.34 (0.18, 0.50)
Normal, n (%) 95/100 (95) 31/47 (66)
FMI index kg/m2b
Reduced, n (%) 0/47 (0) 47/47 (100) 0.61 (0.49, 0.73)
Normal, n (%) 71/100 (71) 29/100 (29)
Men
FFMI kg/m2b
Reduced, n (%) 15/76 (20) 61/76 (80) 0.57 (0.45, 0.69)
Normal, n (%) 71/92 (77) 21/92 (23)
FMI kg/m2b
265
Reduced, n (%) 27/76 (36) 49/76 (64) 0.52 (0.39, 65)
Normal, n (%) 80/92 (87) 12/92 (13)
FFMI = fat-free mass index, FMI = fat mass index, BMI = body mass index. aReduced BMI = <18.5 kg/m2. bLow FFMI for women <14.6 kg/m2 and for men <16.7 kg/m2; low FMI for women <3.9 and for men <1.8 kg/m2.
266
Table 9:4 Baseline characteristics of pulmonary tuberculosis patients with normal versus malnutrition
Normal Low Normal Low
Characteristic BMI BMI FFMI FFMI
(n=437) (n=310) (n=208) (n=103)
n (%) n (%) n (%) n (%)
Sex
Females 236 (54) 116 (37)a 126 (61) 21 (20) a
Males 201 (46) 194 (63) 82 (39) 82 (80)
Age (years)
≤30 257 (59) 177 (57) 135 (65) 59 (57)
>30 1880 (41) 133 (43) 73 (35) 44 (43)
HIV status
Negative 202 (46) 143 (46) 116 (56) 56 (54)
Positive 234 (54) 167 (54) 91 (44) 47 (46)
Hemoglobin (g/dl)1
>10 216 (81) 126 (64)a 151 (76) 73 (74)
267
≤10 51 (19) 70 (36) 48 (24) 26 (26)
Fat mass index (kg/m2)2
Normal 149 (79) 27 (22)a 124 (61) 52 (51)
Low 39 (21) 95 (78) 79 (39) 49 (49)
Chest x-ray disease extent3 75 (17) 36 (12)b 39 (19) Normal/mild 10 (10)b
Moderate/far advanced 362 (83) 274 (88) 169 (81) 93 (90)
Smoker4
No 374 (86) 219 (71)a 176 (85) 62 (60)a
Yes 61 (14) 90 (29) 31 (15) 41 (40)
Takes alcohol5
No 266 (61) 201 (65) 151 (73) 64 (62)b
Yes 170 (39) 109 (35) 56 (27) 39 (38)
History weight loss6
No 111 (26) 54 (17)b 38 (18) 13 (13)
Yes 324 (74) 255 (83) 168 (82) 90 (87) ap-value <0.001, bp-value <0.05, FFMI = fat-free mass index, BMI = body mass index. 1284 missed hemoglobin measurement due to lack of blood; 2FFMI and FFMI was not measured; 3eight missed extent variable; 4three missed history of ever smoked; 5one
268
missed history of alcohol intake; 6four missed history of weight loss; FFMI low <16.7 kg/m2 for men and <14.6 kg/m2 for women, normal ≥ 16.7kg/m2 for men, ≥14.6 kg/m2 for women; low BMI <18.5 kg/m2 for men and women.
269
Table 9:5 Stratified Analysis of mortality among tuberculosis with normal (≥18.5) and low (<18.5) body mass index (BMI kg/m2) according to key variables
[deaths/number at risk (%)]
Normal BMI Low BMI
Category (n=437) (n=310) RR of death (95% CI)
n (%) n (%)
Deaths
No 390 (89) 252 (81) 1.74 (1.22, 2.48)
Yes 47 (11) 58 (19)
Sex
Female 21/236 (9) 17/116 (15) 1.65 (0.90, 3.00)
Male 26/201 (13) 41/194 (21) 1.63 (1.04, 2.56)
Age group
≤30 years 13/257 (5) 21/177 (12) 2.35 (1.21, 4.56)
>30 years 34/180 (19) 37/133 (28) 1.47 (0.98, 2.22)
HIV-serostatus1
Negative 1/202 (0.5) 5/143 (3.5) 7.06 (0.83, 59.81)
270
Positive 46/234 (20) 53/167 (32) 1.61 (1.5, 2.27)
Hemoglobin2
>10 mg/dl 13/216 (6) 17/126 (13) 2.24 (1.13, 4.46)
≤10 mg/dl 10/51 (20) 16/70 (23) 1.17 (0.58, 2.35)
Smoker3
No 41/374 (11) 43/219 (20) 1.79 (1.21, 2.66)
Yes 6/61 (10) 14/90 (16) 1.58 (0.64, 3.89)
Takes alcohol4
No 25/266 (9) 39/201 (19) 2.06 (1.29, 3.30)
Yes 22/170 (13) 19/109 (17) 1.35 (0.77, 2.37)
Weight loss5
No 5/111 (5) 6/54 (11) 2.47 (0.79, 7.72)
Yes 42/324 (13) 52/255 (20) 1.57 (1.08, 2.28)
Chest x-ray extent6
Normal/minimal 9/75 (12) 10/36 (28) 2.31 (1.03, 5.19)
Moderate/far advanced 38/357 (11) 48/271 (18) 1.66 (1.12, 2.47)
1One missed HIV status; 2284 missed hemoglobin measurement due to lack of blood; 6eight missed extent variable; 3three missed history of ever smoked; 4one missed history of alcohol intake; 5four missed history of weight loss; low BMI <18.5 kg/m2.
271
Table 9:6 Stratified Analysis of mortality among tuberculosis with normal and low
fat-free mass index (FFMI kg/m2) according to key variables [deaths/number at risk
(%)]
Normal FFMI Low FFMI
Category (n=208) (n=103) RR of death (95% CI)
Deaths
No 194 (93) 87 (84) 2.31 (1.17, 4.54)
Yes 14 (7) 16 (16)
Sex
Female 6/126 (5) 6/21 (29) 6.00 (2.14, 16.86)
Male 8/82 (10) 10/82 (12) 1.25 (0.52, 3.01)
Age group
≤30 years 5/135 (4) 5/59 (8) 2.29 (0.69, 7.61)
>30 years 9/73 (12) 11/44 (25) 2.03 (0.91, 4.50)
HIV-serostatus1
Negative 0/116 (0) 1/56 (1.8) -
Positive 14/91 (15) 15/47 (32) 2.07 (1.10, 3.92)
272
Hemoglobin2
>10 mg/dl 8/160 (5) 7/77 (9) 1.82 (0.68, 4.83)
≤10 mg/dl 6/48 (13) 9/26 (35) 2.79 (1.11, 6.92)
Smoker3
No 12/176 (7) 11/62 (18) 2.60 (1.21, 5.59)
Yes 2/31 (6) 5/41 (12) 1.89 (0.39, 9.10)
Takes alcohol4
No 10/151 (7) 9/64 (14) 2.12 (0.91, 4.98)
Yes 4/56 (7) 7/39 (18) 2.51 (0.79, 8.00)
Weight loss5
No 2/38 (5) 0/13 (0) -
Yes 12/168 (7) 16/90 (18) 2.49 (1.23, 5.03)
Chest x-ray extent6
Normal/minimal 2/39 (5) 5/10 (50) 9.75 (2.21, 43.06)
Moderate/far advanced 12/164 (7) 11/93 (12) 1.62 (0.74, 3.52)
1One missed HIV status; 215 missed hemoglobin measurement due to lack of blood; 6eight missed extent variable; 3one missed history of ever smoked; 4one missed history of alcohol intake; 5four missed history of weight loss; FFMI low (<16.7 kg/m2 for men and <14.6 kg/m2 for women, normal (≥ 16.7kg/m2 for men, ≥14.6 kg/m2 for women).
273
Table 9:7 Univariate Analysis of factors associated with survival
Characteristic Deaths/N (%) Relative hazard (95% CI)
Fat-free mass index
(kg/m2)a
Normal 14/208 (7) 1
Low 16/103 (16) 2.34 (1.14, 4.80)
Fat mass index (kg/m2)a
Normal 16/176 (9) 1
Low 14/135 (10) 1.23 (0.60, 2.52)
Body mass index (kg/m2)
Normal 47/437 (11) 1
Low 58/310 (19) 1.80 (1.23, 2.64)
Sex
Female 38/352 (11) 1
Male 67/395 (17) 1.66 (1.12, 2.48)
Age (years)
≤30 34/434 (8) 1
274
>30 71/313 (23) 3.15 (2.09, 4.74)
HIV-serostatus1
Negative 6/345 (2) 1
Positive 99/401 (25) 16.24 (7.13, 37.02)
Hemoglobin (g/dl)2
>10 30/342 (9) 1
≤10 26/121 (21) 1.63 (1.00, 2.69)
Smoker3
No 84/593 (14) 1
Yes 20/151 (13) 0.98 (0.60, 1.60)
Takes alcohol4
No 64/467 (14) 1
Yes 41/279 (15) 1.10 (0.75, 1.63)
Chest x-ray extent6
Normal/minimal 19/111 (17) 1
Moderate/far advanced 86/628 (14) 0.73 (0.44, 1.19)
Weight loss5
275
No 11/165 (7) 1
Yes 94/579 (16) 2.34 (1.44, 4.80)
FFMI = fat-free mass index, BMI = body mass index, 95% CI = confidence interval. 1One missed HIV status; 2284 missed hemoglobin measurement due to lack of blood; 2fat and fat-free mass was not measured; 6eight missed extent variable; 3three missed history of ever smoked; 4one missed history of alcohol intake; 5four missed history of weight loss; a311 had FMI and FFMI, respectively); FFMI low <16.7 kg/m2 for men and <14.6 kg/m2 for women, normal ≥ 16.7kg/m2 for men, ≥14.6 kg/m2 for women; low BMI <18.5 kg/m2 for men and women.
276
Table 9:8 Relative hazards [HR, 95% confidence intervals (CIs)] for death among tuberculosis patients with normal compared with patients having low body mass index (BMI)
Overall model Characteristics Deaths/N (%)
HR (95% CI)
BMI (kg/m2)
Normal (≥18.5) 47/437 (11) 1
Low (<18.5) 58/310 (19) 1.85 (1.25, 2.73)
Age (years)
≤30 34/434 (8) 1
>30 71/313 (23) 2.31 (1.53, 3.49)
HIV status1
Negative 6/345 (2) 1
Positive 99/401 (25) 15.86 (6.92, 36.32)
Smoker2
No 84/593 (14) 1
277
Yes 20/151 (13) 0.69 (0.42, 1.14)
Weight loss3
No 11/165 (7) 1
Yes 94/579 (16) 3.40 (1.81, 6.37)
Chest x-ray extent4
Normal/minimal 19/111 (17) 1
Moderate/far advanced 86/628 (14) 0.80 (0.49, 1.32)
1One missed HIV status; 4eight missed extent variable; 2three missed history of ever smoked; 3four missed history of weight loss; low BMI <18.5 kg/m2.
278
Table 9:9 Relative hazards [HR, 95% confidence intervals (CIs)] for death among tuberculosis patients with normal compared with patients having low body mass index (BMI) stratified according to sex status
Stratified models
Deaths/N Characteristics Women (n=352) Men (n=395) (%)
HR (95% CI) HR (95% CI)
BMI (kg/m2)
Normal (≥18.5) 47/437 (11) 1 1
Low (<18.5) 58/310 (19) 1.83 (0.96, 3.50) 1.70 (1.03, 2.81)
Age (years)
≤30 34/434 (8) 1 1
>30 71/313 (23) 1.77 (0.93, 3.37) 2.57 (1.46, 4.54)
HIV status1
Negative 6/345 (2) 1 1
Positive 99/401 (25) 12.61 (3.84, 41.37) 18.48 (5.76, 59.26)
Smoker2
No 84/593 (14) 1 1
279
Yes 20/151 (13) 0.38 (0.05, 2.78) 0.71 (0.37, 1.37)
Weight loss3
No 11/165 (7) 1 1
Yes 94/579 (16) 2.60 (1.00, 6.76) 4.27 (1.84, 9.95)
Chest x-ray extent4
Normal/minimal 19/111 (17) 1 1
Moderate/far advanced 86/628 (14) 0.91 (0.42, 1.99) 0.61 (0.35, 1.04)
1One missed HIV status; 4eight missed extent variable; 2three missed history of ever smoked; 3four missed history of weight loss; low BMI <18.5 kg/m2.
280
Table 9:10 Relative hazards [HR, 95% confidence intervals (CIs)] for death among tuberculosis patients with normal compared with patients having low body mass index (BMI)
Overall model Characteristics Deaths/N (%)
HR (95% CI)
BMI (kg/m2)
Normal (≥18.5) 47/437 (11) 1
Low (<18.5) 58/310 (19) 1.75 (1.18, 2.60)
Age (years)
≤30 34/434 (8) 1
>30 71/313 (23) 3.10 (2.05, 4.69)
Gender
Female 38/352 (11) 1
Male 67/395 (17) 1.63 (1.07, 2.50)
Smoker2
No 84/593 (14) 1
281
Yes 20/151 (13) 0.56 (0.33, 0.93)
Weight loss3
No 11/165 (7) 1
Yes 94/579 (16) 2.70 (1.44, 5.07)
Chest x-ray extent4
Normal/minimal 19/111 (17) 1
Moderate/far advanced 86/628 (14) 0.66 (0.40, 1.09)
1One missed HIV status; 4eight missed extent variable; 2three missed history of ever smoked; 3four missed history of weight loss; low BMI <18.5 kg/m2.
282
Table 9:11 Relative hazards [HR, 95% confidence intervals (CIs)] for death among tuberculosis patients with normal compared with patients having low body mass index (BMI) stratified according to HIV status
Stratified models
HIV negative HIV positive Characteristics Deaths/N (%) (n=402) (n=345)
HR (95% CI) HR (95% CI)
BMI (kg/m2)
Normal (≥18.5) 47/437 (11) 1 1
Low (<18.5) 58/310 (19) 6.95 (0.78, 61.89) 1.63 (1.09, 2.44)
Age (years)
≤30 34/434 (8) 1 1
>30 71/313 (23) 4.89 (0.84, 28.39) 2.10 (1.37, 3.22)
Gender
Female 38/352 (11) 1 1
Male 67/395 (17) 0.57 (0.10, 3.31) 1.62 (1.05, 2.52)
Smoker2
No 84/593 (14) 1 1
283
Yes 20/151 (13) 1.20 (0.18, 8.17) 0.56 (0.33, 0.96)
Weight loss3
No 11/165 (7) 1 1
Yes 94/579 (16) - 3.46 (1.84, 6.49)
Chest x-ray extent4
Normal/minimal 19/111 (17) 1 1
Moderate/far advanced 86/628 (14) - 0.73 (0.44, 1.21)
1One missed HIV status; 4eight missed extent variable; 2three missed history of ever smoked; 3four missed history of weight loss; low BMI <18.5 kg/m2.
284
Table 9:12 Relative hazards [HR, 95% confidence intervals (CIs)] for death among tuberculosis patients with normal compared with low fat-free mass index (FFMI)
Overall model Characteristics Deaths/N (%)
HR (95% CI)
FFMI in kg/m2
Normal (≥18.5) 14/208 (7) 1
Low (<18.5) 16/103 (16) 1.88 (0.96, 3.65)
Age (years)
≤30 11/212 (5) 1
>30 24/128 (19) 2.57 (1.26, 5.28)
HIV-serostatus1
Negative 1/186 (0.5) 1
Positive 34/153 (22) 34.43 (4.66, 254.32)
Hemoglobin (mg/dl)2
>10 18/242 (7) 1
285
≤10 16/83 (47) 1.68 (0.86, 3.28)
1One missed HIV status; 215 missed hemoglobin measurement due to lack of blood; FFMI low (<16.7 kg/m2 for men and <14.6 kg/m2 for women, normal (≥ 16.7kg/m2 for men, ≥14.6 kg/m2 for women).
286
Table 9:13 Relative hazards [HR, 95% confidence intervals (CIs)] for death among
tuberculosis patients with normal compared with low fat-free mass index (FFMI) stratified according to sex status
Stratified models
Characteristics Deaths/N (%) Women (n=165) Men (n=175)
HR (95% CI) HR (95% CI)
FFMI in kg/m2
Normal (≥18.5) 14/208 (7) 1 1
Low (<18.5) 16/103 (16) 6.83 (2.14, 21.74) 0.80 (0.35, 1.84)
Age (years)
≤30 11/212 (5) 1 1
>30 24/128 (19) 3.54 (1.09, 11.47) 2.22 (0.87, 5.66)
HIV-serostatus1
Negative 1/186 (0.5) 1 1
Positive 34/153 (22) - 21.44 (2.84, 161.99)
Hemoglobin (mg/dl)2
>10 18/242 (7) 1 1
287
≤10 16/83 (47) 0.73 (0.23, 2.32) 2.88 (1.26, 6.57)
1One missed HIV status; 215 missed hemoglobin measurement due to lack of blood; FFMI low (<16.7 kg/m2 for men and <14.6 kg/m2 for women, normal (≥ 16.7kg/m2 for men, ≥14.6 kg/m2 for women).
288
Figure 9:1 Survival distribution among adult men presenting with wasted body mass (BMI <18.5 kg/m2) compared to men with normal body mass in urban
Kampala, Uganda
289
Figure 9:2 Survival distribution among adult women with baseline fat-free mass wasting (FFMI <16.7 kg/m2 for men, <14.6 kg/m2 for women) compared to women with normal fat-free mass in urban Uganda
290
Figure 9:3 Survival distribution Men compared with Women among HIV positive tuberculosis patients in urban Uganda
Data truncated at 1462 days of follow-up
291
CHAPTER 10
LONGITUDINAL CHANGES IN BODY COMPOSITION AMONG HIV
POSITIVE AND HIV NEGATIVE ADULT PATIENTS WITH PULMONARY
TUBERCULOSIS IN URBAN KAMPALA, UGANDA
292
Abstract
Background Follow-up studies are limited to understand nutritional changes during and after tuberculosis treatment. The effect of body wasting at the time of diagnosis and HIV infection on rate of change for fat-free mass, fat mass, and body mass index (BMI) during and after treatment were evaluated among tuberculosis patients in urban Kampala,
Uganda.
Methods In a retrospective cohort study of 717 adult patients, BMI was assessed at baseline, 2, 3, 5, 6, 12, and 24 months whereas fat-free mass index (FFMI) and fat mass index (FMI) were evaluated at baseline, 3, 12, and 24 months. Longitudinal mixed effects two piecewise models with knots at month 3 and month 12 were fit to the data.
Results There were no differences in body wasting as assessed by reduced FFMI, FMI, and BMI between HIV positive and HIV negative patients at presentation. In stratified mixed effects two spline models during the first three months of treatment, the gain in
FFMI among patients that presented with reduced fat-free mass at diagnosis was dramatic in men with rate of 4.55 kg/m2 (95% confidence interval (CI): 1.26, 7.83); however, the
gain was minimal among women who presented with reduced FFMI with rate of 2.07
kg/m2 (95% CI: -0.74, 4.88). In stratified models for FMI as dependent variable, women
with reduced fat mass at presentation had a substantial gain in FMI at rate of 3.55 kg/m2
(95% CI: 0.40, 6.70) whereas men had a rate of 3.16 kg/m2 (0.80, 5.52). In stratified
models with BMI as dependent variable, men with reduced BMI at presentation gained
BMI at a rate of 6.45 kg/m2 (95% CI: 3.02, 9.87) whereas women at a rate of 3.30 kg/m2
(95% CI: -0.11, 6.72). There were minimal changes in FFMI, FMI, and BMI during the
293
first three months of treatment in stratified models according to HIV status. Furthermore, there were minimal changes in FFMI, FMI, and BMI after month 3 and during the one year follow-up after month 12.
Conclusion Gender but not HIV was associated with longitudinal body composition changes during the initial phase of treatment among tuberculosis patients that presented with body wasting. Further evaluation is needed to under the impact of providing nutritional interventions as adjuvant treatment on body composition among tuberculosis patients in sub-Saharan.
294
Background
The global burden of tuberculosis remains enormous because there are insufficient
tuberculosis control programs and higher rates of tuberculosis and human
immunodeficiency virus (HIV) co-infection. The incidence remains high, with 9.2 million new cases reported to have occurred in 2006 only (Vitoria et al. 2009). The majority of these cases occurred in Asia, but the highest population rate has been observed in Africa where the prevalence of HIV infection is at its highest.
Body wasting is regarded as a cardinal feature of tuberculosis; however, the pathophysiology of wasting remains poorly understood (Schwenk A and Macallan D.C
2000). A significant proportion of African tuberculosis patients have a marked degree of wasting by the time they present for registration and treatment (Kennedy et al. 1996;
Harries et al. 1988; Zachariah et al. 2002). Wasting is associated with impaired physical function (Harries et al. 1988), longer hospitalization days and increased mortality in patients with tuberculosis (Rao et al. 1998; Zachariah et al. 2002; Mitnick et al. 2003).
Thus, early diagnosis of body wasting in tuberculosis is essential if timely interventions are to be instituted.
Wasting associated with tuberculosis is likely caused by a combination of decreased appetite and altered metabolism resulting from the inflammatory and immune responses
295
(Paton et al. 1999; Paton et al. 2003; Macallan 1999). The decreased appetite may lead to
a decrease in energy intake, interacting with increased losses with a resultant body
wasting. Although antituberculosis treatment is highly successful (Chemotherapy and
management of tuberculosis in the United Kingdom: recommendations 1998. Joint
Tuberculosis Committee of the British Thoracic Society 1998), many patients remain
underweight after 6 months of treatment (Onwubalili 1988) suggesting that full recovery may take a longer than the treatment itself.
It has been reported that in patients with both tuberculosis and HIV co-infection, the
wasting process is exacerbated (Macallan 1999; Lucas et al. 1994). In contrast, findings
from several cross-sectional studies appear to show no large differences in body
composition between HIV-infected adults with tuberculosis and HIV-negative adults with
tuberculosis at presentation (Shah et al. 2001; Niyongabo et al. 1999; Niyongabo et al.
1994; Mupere et al. 2010) suggesting that tuberculosis is the dominant factor inducing
wasting. This has been shown in several reports (Mupere et al. 2010; Paton and Ng
2006). However, the wasting process at the time of TB diagnosis according to previous
reports appear to differ by gender (Kennedy et al. 1996; Mupere et al. 2010).
Nutritional changes at the time of tuberculosis diagnosis are well documented using both
anthropometry and bioelectrical impedance analysis (BIA) in several reports (van Lettow,
Fawzi, and Semba 2003; Paton et al. 1999; Shah et al. 2001; Van Lettow et al. 2004;
Villamor et al. 2006; Mupere et al. 2010); however, follow-up studies are still limited to
296
understand nutritional changes during and after tuberculosis treatment. The limited follow-up studies (Kennedy et al. 1996; Ramakrishnan et al. 1961) found weight gain to be a poor indicator of clinical response in tuberculosis. Furthermore, weight or body mass index (BMI) provides limited information about nutritional alterations in patients with tuberculosis. In addition, BMI is insensitive to body fatness, particularly at low BMI, as well as with above-normal muscle development (Kyle, Genton, and Pichard 2002; Kyle,
Piccoli, and Pichard 2003). Measurement of body composition is necessary to obtain a true picture of the nutritional status in tuberculosis because body compartments differ in their contribution to weigh gain and its clinical benefit. Fat-free mass (FFM) and fat mass body composition measurements have been shown to permit a more precise evaluation of nutritional status (VanItallie et al. 1990; Kyle, Piccoli, and Pichard 2003). Fat-free mass
(FFM) is more closely correlated with quality of life and physical functioning than are fat mass (FM) and body weight (Wagner, Ferrando, and Rabkin 2000; Mostert et al. 2000).
BIA has been recommended as the preferred and precise method for clinical assessment of FFM and fat mass (Kyle et al. 2004; Kyle, Genton, and Pichard 2002). One follow-up study that evaluated FFM and FM had a small sample size, enrolled only HIV negative tuberculosis patients, and followed patients for only 6 months (Schwenk et al. 2004). In this large retrospective study, we evaluated the effect of baseline body wasting and HIV infection on the rate of change of FFM, FM, and BMI during and after treatment among tuberculosis patients in urban Uganda, Kampala.
297
Methodology
Study Design
A retrospective cohort study was conducted with the study population that consisted of
745 adult pulmonary tuberculosis patients having confirmed HIV status and defined baseline body wasting. The study used the completed five year Household Contact
(HHC) study, the completed phase II prednisolone double blind randomized placebo controlled clinical trial, and the ongoing Kawempe Community Health (KCH) study. Of the 745 patients, 314 were enrolled into the HHC, 341 into the KCH, and 90 into the placebo arm of the prednisolone clinical trial (Figure 10:1 and 10:2). The HHC and KCH studies were observational epidemiologic studies; organized and conducted by the
Makerere University and Case Western Reserve University tuberculosis research collaboration (Uganda-CWRU) that has been ongoing for the last 20 years in Uganda.
The HHC was the initial household contact study from 1995 to 1999 that described the epidemiology of tuberculosis in urban Kampala, Uganda (Guwatudde et al. 2003). The
KCH is the second phase of the HHC. The KCH phase started in 2002 and is still ongoing
(Stein et al. 2005). The KCH was developed specifically to focus on the determinants of host factors associated with primary infection, re-infection, reactivation, and progression of clinical disease and to identify and track individual strains of mycobacterial tuberculosis through Ugandan households and local community. The phase II clinical trial was conducted between 1995 to 2000 to determine whether immunoadjuvant prednisolone therapy in HIV-infected patients with TB who had CD4(+) T cell counts
>/=200 cells/ mu L was safe and effective at increasing CD4(+) T cell counts (Mayanja-
Kizza et al. 2005). Patients were eligible for this present study if they had baseline and
298
follow-up body measurements, had an HIV test, and were part of one prospective study
conducted by the Uganda-CWRU research collaboration.
The institutional review boards at Case Western Reserve University in the United States
and Joint Clinical Research Center in Uganda reviewed the protocol and final approval
was obtained from the Uganda National Council for Science and Technology. All
patients in the HHC and KCH had written informed consent to be enrolled in the study.
All participants in both HHC and KCH were given appropriate pre- and post-test HIV counseling and AIDS education. HIV-1 infection was diagnosed on the basis of a positive enzyme-linked immunosorbent assay for HIV-1 antibodies (Recombigen; Cambridge
Biotech, Cambridge, MA). None of the HIV positive patients, neither those who were newly identified with HIV nor those with pre-existing HIV, were on antiretroviral therapy.
At enrollment, basic demographic information and a medical history were collected, and a standardized physical examination was conducted by a medical officer. Active tuberculosis was confirmed by sputum smear microscopy and culture. Patients with active tuberculosis were treated with standard four-drug chemotherapy for tuberculosis per guidelines of the Ugandan Ministry of Health. Adults with a previous history of treated pulmonary tuberculosis were excluded in the study. Of the 745 participants who were enrolled in the three studies, 28 were excluded due to lack of anthropometric follow-up data and being below 18 years of age, leaving 717 participants in total
299
available for analysis. The BIA data were collected during the KCH study only. Of the
341 participants who were enrolled in KCH, 63 were excluded due to lack of BIA follow-
up data and being below 18 years of age, leaving 278 participants in total available for
analysis (Figure 10:2). However, there were no differences in baseline age, gender,
weight, height, BMI, smoking status, hemoglobin, chest x-ray disease extent, and history
of weight loss between participants who were included and those who were excluded.
Measurements
Nutritional status was assessed using anthropometric measurements such as height and
weight and BIA Detroit, MI, RJL Systems. Body weight was determined to the nearest
0.1 kg using a SECA adult balance, and standing height was determined to the nearest 2 mm. Anthropometric measurements for the present retrospective study were performed during the HHC and KCH at scheduled visits including baseline, 2, 3, 5, 6, 12, and 24 months on follow-up. Body-mass index (BMI) was computed using the relationship of weight in kilograms divided by height in meters squared (kg/m2). All BIA measurements
were performed by one trained observer using the same equipment and recommended
standard conditions with regard to body position, previous exercise, dietary intake, skin
temperature, and voiding of the bladder were taken into consideration in taking BIA
measurements (Kyle et al. 2004). All BIA measurements during the KCH study were
performed on the day patients were confirmed to have tuberculosis disease and on
scheduled visits at 3, 12, and 24 months.
300
The BIA is a simple, easy, safe, non-invasive technique, that has been recommended for nutritional studies in the clinical setting (Kyle et al. 2004; Kyle, Piccoli, and Pichard
2003) and is a convenient method to determine the lean or fat-free mass and fat body compartments (Kyle, Piccoli, and Pichard 2003; Kyle et al. 2004). Single-frequency BIA was performed at 50 kHz and 800 mA with standard tetrapolar lead placement (Jackson et al. 1988). Before performing measurements on each participant, the BIA instrument was calibrated using the manufacturer’s recalibration device. The resistance and reactance were based on measures of a series circuit (Kotler et al. 1996). BIA measurements were performed in triplicate for each subject. Fat-free mass (FFM) was calculated from BIA measurements using equations that were previously cross-validated in a sample of patients (white, black and Hispanic) with and without HIV infection
(Kotler et al. 1996) and have been applied elsewhere in African studies (Villamor et al.
2006; Shah et al. 2001; Van Lettow et al. 2004). Fat mass (FM) was calculated as body weight minus FFM.
Definitions
We used BMI and height-normalized indices (adjusted for height2) of body composition that partition BMI into fat-free mass index (FFMI) and fat mass index (FMI) (Schutz,
Kyle, and Pichard 2002; VanItallie et al. 1990; Kyle, Piccoli, and Pichard 2003) to establish the body wasting status of participants. The FFMI and FMI have the advantages of compensating for differences in height and age (Kyle, Genton, and Pichard 2002).
Also, the use of the FFMI and FMI eliminates some of the differences between
301
population groups. We defined body wasting as patients having the low fat-free mass
index (FFMI) and the low body fat mass index (FMI) corresponding to WHO BMI
categories for malnutrition as previously reported (Table 10:1) (Kyle, Piccoli, and
Pichard 2003). The FFMI <16.7 (kg/m2) for men and <14.6 (kg/m2) for women and the
FMI <1.8 (kg/m2) for men and <3.9 (kg/m2) for women corresponds to a BMI of <18.5
kg/m2, the WHO cutoff for malnutrition (WHO Tech Rep 1995) among adults.
Table 10:1 Definitions of low and normal fat and fat-free mass index values for corresponding body mass index in adults
Characteristic Low Normal
Body mass index (BMI)a
Women and men in kg/m2 < 18.5 ≥ 18.5
Fat-free mass index (FFMI)b
Women in kg/m2 < 16.7 ≥ 16.7
Men in kg/m2 < 14.6 ≥ 14.6
Fat mass index (FMI)b
Women in kg/m2 < 1.8 ≥ 1.8
Men in kg/m2 < 3.9 ≥ 3.9
aWorld Health Organization categories, sex independent (WHO Tech Rep 1995). bKyle et al (Kyle, Piccoli, and Pichard 2003; Kyle et al. 2003).
302
Statistical analysis
Hypothesis
The hypothesis tested in analysis of this data was: the null hypothesis stated that there would be no difference in the rate of change for FFMI, FMI, and BMI between patients with baseline body wasting and patients with no baseline wasting; and between HIV positive and HIV negative patients during and after the course of tuberculosis treatment in urban Kampala, Uganda. The study hypothesis (alternative) was that patients with baseline wasting and HIV infection were associated with increased rates of FFMI, FMI, and BMI compared to patients with no baseline wasting and with HIV negative during and after treatment among tuberculosis patients in urban Kampala, Uganda.
Descriptive statistics
Baseline characteristics for participants with baseline wasting were compared with participants who had no baseline wasting using the χ2 test or Fisher’s exact test (where
expected counts were less than 5) for binary data and student’s t-test for continuous
variables or Wilcoxon-Mann Whitney test for variables not normally distributed.
Exploration of data
In order to gain insights for the covariate structure and model building, we explored data
by conducting correlation structure matrices for FFMI, FMI, and BMI; plots of individual
profile trajectories, boxplots, and mean profiles per exposure group of baseline wasting
303
and no wasting overtime. The raw correlation coefficients along and off the diagonals
were clearly unique for BMI suggesting a heterogeneous covariance structure. However,
correlation coefficients for both FFMI and FMI were nearly the same along the diagonals
and appear to decay with time suggesting stationary structure (Appendix, Table 14:1).
Plots of individual FFMI, FMI, and BMI profiles over time have different intercepts and
different positive slopes suggesting importance of random intercepts and slopes in the
models (Appendix, Figures 14:1 – 14:3). The boxplots and mean profiles for three indexes showed linear increasing trends over time suggesting inclusion of interaction terms between the indexes with time to be important (Appendix, Figures 14:4 – 14:15).
Missing observations
The 278 participants from the KCH study who were included in the BIA analysis contributed 1,112 observations, 205/1,112 (18%) of which were missing observations.
However, the 717 participants from all the three studies who were included in the anthropometric data analysis contributed 4,875 observations, 781/4,875 (16%) of which were missing observations (Figure 10:1). In order to understand the nature of missingness and whether the missing observations were nonignorable, we used the logistic regression approach for outcome-dependent missing (Ridout 1991). We performed separate generalizing estimating equations (GEE) models for FFM, FMI, and BMI using Proc
GENMOD with a logit link function to assess the effect of having wasting (low FFMI, low FMI, or BMI <18.5 kg/m2) versus no wasting (normal/high FFMI, normal/high FMI,
or BMI ≥18.5 kg/m2) at baseline on to the probability of missing FFM, FMI, or BMI
304
overtime adjusting factor for baseline sex, age, HIV status, history of weight loss, chest x-ray disease extent, and hemoglobin level. In both models of FFMI and FMI, for outcome dependent missing, the missing probability was not related to the baseline wasting and the lag measure (Appendix, Tables 14:2 and 14:3). Thus, we assumed the data missing not to be informative on the outcome variable and ignorable for FFMI, FMI and BMI.
Model building
The aim was to specify the appropriate mean and covariance structures of the data. The main outcome vectors of the participant’s FFMI, FMI, and BMI were modeled as linear combinations of baseline covariates and baseline body wasting status in separate models.
Changes in FFMI, FMI, or BMI over time were estimated using multilevel linear random effect/mixed effects models with random intercept and random slope, accounting for correlation of repeated measurements within each individual (Laird and Ware 1982;
Singer J.D and Willet J.B 2003; Bryk A.S and Ruandenbush S.W 2002). Multilevel modeling of change involves two levels: level one that describes how the outcome in each individual changes overtime (within-person change), and level two that describes how the within-person changes differ across persons (between-person differences in change) (Singer J.D and Willet J.B 2003). We used full maximum likelihood estimation method for parameter estimation and type 3 F-test for testing significance. We assumed a spatial exponential covariance structure. The spatial exponential covariance structure was plausible because the intervals between serial data points were different by design. The
305
exponential covariance structure is among the spatial correlation structures that takes the form: εij = Ui(tij) + еij. The Ui(tij) are assumed to have a normal distribution, with 0 mean,
(|tij – tik|) variance σ2 u, and correlation Corr { Ui(tij), Ui(tik)} = ρ . This correlation becomes
weaker as the separation increases. The еij are the usual sampling or measurement errors.
The LOCAL option was added to the REPEATED statement in SAS to decompose the
errors into Ui(tij) and еij.
The appropriateness of the covariance structure was assessed using Akaike Information
Criteria (AIC) and Bayesian Information Criteria (BIC) (Littell, Pendergast, and
Natarajan 2000). The quadratic and cubic terms were assessed using –2 log likelihood (-
2LL) and found to significantly improve the model (Appendix, Table 14:4). A model
with random intercepts and random slopes was assessed using -2LL and found to be
better than one without random intercepts or random slopes (Appendix, Table 14:4). We
also compared a piecewise model with two knots at month 3 and at month 12 with linear
and polynomial models. The knots in the piecewise were chosen a priori basing on the
scheduled data point measurements. The piecewise model was found to fit the data better
(Appendix, Table 14:5). Furthermore, a piecewise model with random intercepts and
random slopes was better than one without. We assessed for normality of the residuals at
modeling levels to identify the best function form of each variable included in the model.
We first fitted an unconditional mean model (outcome only, no predictors) to determine
whether the differences in average FFMI, FMI, and BMI across visits and persons were
306
non-zero and whether there was significant within- and between-person variability
(Singer J.D and Willet J.B 2003). After confirming non-zero variability in FFMI, FMI,
and BMI across visits and persons, and the existence of significant within-person and
between person variability, an unconditional growth model (with time as the only
predictor) was then fit to estimate the population average rate of FFMI, FMI, BMI
increase (slope) and initial status (intercept), and to determine if significant variability
existed in both statistics. The unconditional growth model showed significant variability
in both the initial status and the rate of change of FFMI, FMI, and BMI, suggesting the
need for predictor variables to explain the heterogeneity. The main predictor was baseline
body wasting using the following cutoffs alluded to above: low FFMI of 16.7 (kg/m2) for
men and 14.6 (kg/m2) for women, low FMI of 1.8 (kg/m2) for men and 3.9 (kg/m2) for
women (Kyle, Piccoli, and Pichard 2003; Schutz, Kyle, and Pichard 2002), and the
corresponding BMI cutoff of <18.5 kg/m2 (WHO Tech Rep 1995). The following were
other covariates used to assess associations with the dependent variables overtime:
number of visits in the study to adjust for missing data points, age categorized as young
(≤ 30 years) and old (> 30 years), sex, HIV sero-status, hemoglobin categorized as
anemic (≤ 10g/dl) and not anemic (>10), smoking status, history of weight loss, history of
taking alcohol now, and chest x-ray disease extent categorized as normal/minimal and moderate/far advanced. All covariates were time-invariant. The interval between study visits was adjusted to an annual scale to facilitate the interpretation of rates of FMI, FMI, and BMI decline per year.
307
We performed univariate analysis; variables with significant linear rates of FFMI, FMI, and BMI change and those with biological association with the three indexes such as age and extent of chest x-ray disease involvement were included in the initial multivariable model. Conditional mixed-effects model was fitted to the BMI data in view of the informative missing data (non-ignorable). Age was the only variable that was dropped from all the final multivariable models because it did not improve the model fits. The complete set of the covariates at each visit were used in the analysis; neither last- observed-carried-forward (LOCF) nor imputation was applied. Differences in rates of
FFMI, FMI, and BMI change were assessed by fitting in separate models with baseline body wasting for FFMI, FMI, and BMI cutoffs*time interaction term variables. To assess for the confounding effects of HIV, we fitted stratified models according to HIV strata.
Follow-up data were censored at the last visit. We assessed for the basic assumptions of multivariate normality and linearity, all final models had normally distributed residuals.
The analysis will be performed using SAS MIXED procedures (SAS Institute). All analyses were performed using SAS version 9.1.3 Cary software, North Carolina SAS
Institute Inc. 2004.
Results
Descriptive statistics
Of the 717 patients who were included in the analysis, 293 had reduced BMI (<18.5 kg/m2 for women and men) at diagnosis (Figure 10:1). Among the 293 with reduced
308
BMI, 153 were HIV positive and 115 were women. Two hundred seventy eight patients,
a subset of the 717 were involved in the fat and fat-free mass analysis (Figure 10:2). Of
the 278 patients, 94 had reduced FFMI (<16.7 kg/m2 for men and <14.6 kg/m2 for
women). Among the 94 with reduced FFMI, 41 were HIV positive and 19 were women.
Of the 278 patients, 120 had reduced FMI (<1.8 kg/m2 for men and <3.9 kg/m2 for
women). Among the 120 with reduced FMI, 52 were HIV positive and 69 were women.
Baseline characteristics of the study population are presented in Table 10:2. There were
gender differences at the time of diagnosis. Men had significantly higher proportions of
reduced FFMI and BMI compared to women whereas women had significantly higher
proportions of reduced FMI compared to men. There were no differences in body wasting
between HIV positive and HIV negative patients at presentation regardless of nutritional
assessment measure suggesting that HIV probably does not affect body wasting in co- infected patients as previously reported (Mupere et al. 2010; Paton and Ng 2006).
Patients who presented with history of prior smoking and moderate/or far advanced disease extent on chest x-ray had higher frequency of body wasting as assessed by FFMI and BMI compared to those without history of prior smoking and without moderate/or far advanced disease on chest x-ray (Table 10:2).
309
Rate of change for fat-free mass, fat mass, and body mass index in univariate spline
models
Fat-free mass index
Overall, the population average FFMI linear trajectory had a rate of 2.43 (95% CI, 1.34,
3.52) kg/m2 increase per month before month 3 on tuberculosis treatment in univariate mixed effects model with FFMI as the dependent variable (Table 10:3). However, after
month 3 the FFMI decreased significantly at a rate of -2.57 (95% CI; -3.87, -1.27) kg/m2
per month. There were minimal changes in FFMI after month 12 over the subsequent one
year. On average at baseline, month 3, and month 12, FFMI was -1.40 (95% CI; -1.69, -
1.11), -0.76 (95% CI; -1.17, -0.35), and -1.49 (95% CI; -2.12, -0.85) kg/m2 significantly
lower among patients that presented with reduced fat-free mass compared to patients that had normal fat-free mass, respectively. Patients that presented with reduced fat-free mass
had a significantly rapid gain in FFMI before 3 months on treatment at rate of 4.02 (95%
CI; 3.14, 4.91) kg/m2 compared to a modest gain of 1.59 (95% CI; 0.96, 2.23) kg/m2 per
month among those that had normal fat-free mass (Table 10:3). The difference in slopes
was 2.43 (95% CI; 1.34, 3.52). Before month 3 is the intensive phase of tuberculosis
treatment. There were no gains in FFMI after month 3 regardless of the amount of fat-
free mass a patient may have presented with at diagnosis. This period coincides with
continuous phase of tuberculosis treatment. However after month 12, patients presenting
with reduced fat-free mass maintained the significant rapid gain compared to patients that
had normal fat-free mass as suggested by the significant difference in slopes of 2.52
(95% CI; 1.17, 3.87) after 12 months (Table 10:3).
310
Fat mass index
In univariate mixed effects spline model for FMI (Table 10:3), there was a significant
gain in the average rate of FMI linear trajectory for the overall population before month 3
at 2.41 (95% CI: 1.33, 3.48) per month. After month 3, the rate of linear trajectory
decreased at a rate of -2.46 (95% CI: -3.78, -1.13) per month. The rate of linear trajectory was minimal per year after month 12. On average, FMI was -2.28 (95% CI; -2.76, -1.80),
-1.67 (95% CI; -2.18, -1.16), and -2.50 (95% CI; -3.37, -1.62) kg/m2 significantly lower
among patients presenting with reduced fat mass compared to patients presenting with
normal fat mass at baseline, month 3, and month 12, respectively. Patients presenting
with reduced fat mass had a significantly rapid gain in FMI before month 3 at rate of 2.99
(95% CI; 2.18, 3.80) kg/m2 compared to a no significant gain of 0.58 (95% CI; -0.12,
1.29) kg/m2 per month among those presenting with normal fat mass at diagnosis (Table
10:3). There was significant difference in slopes of 2.41 (95% CI; 1.33, 3.48). Both
patients with reduced or normal fat mass at presentation gained in FMI at the same rate
per month after month 3. However after month 12, only patients presenting with reduced
fat mass attained the significant gain per year compared to patients that presented with
normal fat mass with a difference of 2.62 (95% CI: 2.22, 4.03) in slopes (Table 10:3).
Body mass index
Unadjusted rate of BMI changes in mixed effects spline model with BMI as the
dependent variable are presented in Table 10:3. The overall population average rate of
BMI linear trajectory gain was 2.72 (95% CI, 1.74, 3.69) kg/m2 per month before month
311
3. However, after month 3 the BMI decreased significantly at a rate of -2.94 (95% CI; -
4.07, -1.81) kg/m2 per month. There was minimal change in the total population rate of
BMI change after month 12. On average at baseline and at month 12, population BMI was 2.72 (95% CI; 2.74, 3.69) and 2.23 (95% CI; 1.61, 4.05) kg/m2 significantly lower among patients that presented with reduced BMI compared to patients that presented with normal BMI, respectively. There was minimal difference in population BMI at month 3 between patients that presented with reduced BMI and normal at diagnosis. Patients that presented with reduced BMI had a significantly higher rate of gain in BMI before 3 months and after month 12 compared to the rate among patients that presented with normal BMI (Table 10:3). The differences in slopes were 2.72 (95% CI; 1.74, 3.69) before month and 2.83 (95% CI; 1.71, 4.05) after month 12, respectively. There was no difference in slope for rate of BMI gain between patients had and patients that did not have reduced BMI at presentation after month 3.
Rate of change for fat-free mass, fat mass, and body mass index in multivariable mixed spline models and stratification models according to gender
Fat-free mass
In multivariable mixed effects two spline model at month 3 and at month 12 with FFMI as the dependent variable and after adjusting for HIV, anemia (hemoglobin ≤10 mg/dl), prior smoking status, history of weight loss, and extent of disease on chest x-ray, the population average for FFMI at baseline, month 3, and month 12 was -1.60 (95% CI: -
1.90, -1.31), -1.02 (95% CI: -1.40, -0.64), and -1.10 (95% CI: -1.50, -0.70) kg/m2
312
significantly lower among patients that presented with reduced fat-free mass compared to patients that had normal fat-free mass, respectively. On average among women at
baseline, month 3, and month 12, FFMI was -1.85 (95% CI: -2.29, -1.40), -1.27 (95% CI:
-1.79, -0.76), and -1.27 (95% CI: -1.80, -0.73)] significant lower among individuals that presented with reduced fat-free mass compared to those that had normal fat-free mass at presentation, respectively. Similarly, among men at baseline, month 3, and month 12,
FFMI was -2.15 (95% CI; -2.45, -1.85), -1.68 (95% CI: -2.11, -1.24), and -1.89 (95% CI:
-2.35, -1.43) for individuals that presented with reduced fat-free mass compared to patients that had normal fat-free mass, respectively.
In the overall two spline model during the first three months, patients that presented with reduced fat-free mass had a significantly gain in FFMI at rate of 2.68 (95% CI: 0.68,
4.67) kg/m2 per month whereas patients that presented with normal fat-free mass had minimal change in FFMI at a rate of 0.35 (95% CI: -1.38, 2.08) kg/m2 per month (Table
10:4). In stratified mixed effects two spline models, the gain in FFMI among patients that
presented with reduced fat-free mass at diagnosis was however dramatic in men at rate of
4.55 (95% CI: 1.26, 7.83) (Table 10:5). The gain was minimal among women that
presented with reduced fat-free mass with a rate of 2.07 (95% CI: -0.74, 4.88). The
slopes were significantly different during the first three months between patients that
presented with reduced fat-free mass and those that had normal fat-free mass at diagnosis
for the overall population model and in stratified models for women and men. There were
no changes in FFMI after month 3 and during the one year of follow-up after month 12 in
the total population and in stratified models for women and men (Tables 10:4 and 10:5).
313
Fat mass
After adjusting for HIV, anemia (hemoglobin ≤10 mg/dl), prior smoking status, history of weight loss, and extent of disease on chest x-ray in multivariable mixed effects two spline model at month 3 and at month 12 with FMI as the dependent variable, the total population average for FMI at baseline, month 3, and month 12, was -2.15 (95% CI: -
2.62, -1.69), -1.57 (95% CI: -2.06, -1.08), and -1.64 (95% CI: -2.24, -1.04)] significant lower among patients that presented with reduced fat mass compared to those that had normal fat mass at diagnosis, respectively. On average among women at baseline, month
3, and month 12, FMI was -4.02 (95% CI: -4.65, -3.39), -3.17 (95% CI: -3.90, -2.58), and
-3.46 (95% CI: -4.34, -2.37) significant lower among individuals that presented with reduced fat mass compared to those that had normal fat mass at diagnosis, respectively.
Similarly, among men at baseline, month 3, and month 12, FMI was -1.28 (95% CI; -
1.47, -1.09), -0.76 (95% CI: -1.09, -0.44), and -0.88 (95% CI: -1.26, -0.50) for individuals that presented with reduced fat mass compared to patients that had normal fat mass, respectively.
During the first three months in the overall two spline model, patients that presented with reduced fat mass had a significantly gain in FMI at rate of 2.23 (95% CI: 0.30, 4.16) kg/m2 per month whereas patients that presented with normal fat mass had minimal
change in FMI at a rate of -0.11 (95% CI: -1.98, 1.73) kg/m2 per month (Table 10:4). In
stratified mixed effects two spline models, the gain in FMI among patients that presented
with reduced fat mass at diagnosis was higher in women at rate of 3.55 (95% CI: 0.40,
314
6.70) compared to men at rate of 3.16 (95% CI: 0.80, 5.52) (Table 10:5). The slopes were
significantly different during the first three months between patients that presented with
reduced fat mass and those that had normal fat mass at diagnosis for the overall
population model and in stratified models for women and men. After month 12 during the
one year follow-up, both patients that presented with/ or without reduced fat mass had
modest significant gains in fat mass; however, there was no significant difference in
slopes (Tables 10:4 and 10:5). There were minimal changes in FMI after month 3 and
before month 12 among women and men regardless of initial fat mass.
Body mass index
The population average BMI at baseline, month 3, and month 12 in multivariable mixed
effects two spline model at month 3 and at month 12 with BMI as the dependent variable
and after adjusting for HIV, anemia (hemoglobin ≤10 mg/dl), prior smoking status,
history of weight loss, and extent of disease on chest x-ray was -3.69 (95% CI; -4.09, -
3.30), -3.30 (95% CI; -3.71, -2.88), and -3.62 (95% CI; -4.16, -3.07) kg/m2 significantly
lower among patients that presented with reduced BMI compared to patients that had normal BMI, respectively. On average among women at baseline, month 3, and month
12, BMI was -4.23 (95% CI: -4.97, -3.48), -3.89 (95% CI: -4.67, -3.12), and -3.86 (95%
CI: -4.88, -2.84)] significant lower among individuals that presented with reduced BMI compared to those that had normal BMI at diagnosis, respectively. Similarly, among men at baseline, month 3, and month 12, BMI was -3.13 (95% CI; -3.49, -2.77), -2.71 (95%
315
CI: -3.13, -2.29), and -3.19 (95% CI: -3.71, -2.66) for individuals that presented with reduced BMI compared to individuals that had normal BMI, respectively.
In the overall two spline model during the first three months, patients that presented with
reduced BMI had a significantly gain in BMI at rate of 3.83 (95% CI: 1.67, 5.99) kg/m2
per month whereas patients that presented with normal BMI had modest gain in BMI at a
rate of 2.25 (95% CI: 0.33, 4.17) kg/m2 per month (Table 10:4). In stratified mixed
effects two spline models, the gain in BMI among patients that presented with reduced
BMI at diagnosis was however substantial in men at rate of 6.45 (95% CI: 3.02, 9.87)
(Table 10:5). The gain was minimal among women that presented with reduced BMI with
a rate of 3.30 (95% CI: -0.11, 6.72). The slopes were significantly different during the
first three months between patients that presented with reduced BMI and those that had
normal BMI at diagnosis for the overall population model and in stratified models for
men. There were minimal changes in BMI after month 3 and during the one year of
follow-up after month 12 in the total population and in stratified models for women and
men (Tables 10:4 and 10:5).
316
Rate of change for fat-free mass, fat mass, and body mass index in multivariable
mixed spline models and stratification models according to HIV status
Fat-free mass
In multivariable mixed effects two spline model at month 3 and at month 12 with FFMI
as the dependent variable when HIV was replaced by gender among the adjusters (Tables
10:6 and 10:7), the population average for FFMI at baseline, month 3, and month 12 was
-2.05 (95% CI: -2.31, -1.80), -1.57 (95% CI: -1.90, -1.24), and -1.72 (95% CI: -2.06, -
1.37) kg/m2 significantly lower among patients that presented with reduced fat-free mass compared to patients that had normal fat-free mass, respectively. On average among HIV negative individuals in stratified models at baseline, month 3, and month 12, FFMI was -
1.77 (95% CI: -2.04, -1.49), -1.35 (95% CI: -1.78, -0.93), and -1.51 (95% CI: -1.93, -
1.10)] significant lower among individuals that presented with reduced fat-free mass compared to those that had normal fat-free mass at presentation, respectively. Similarly, among HIV positive individuals at baseline, month 3, and month 12, FFMI was -2.40
(95% CI; -2.83, -1.98), -1.80 (95% CI: -2.32, -1.28), and -1.98 (95% CI: -2.51, -1.33) significantly lower for individuals that presented with reduced fat-free mass compared to individuals that had normal fat-free mass, respectively.
In the overall model and in stratified models according to HIV status during the three months, after month 3, and during the one year period of follow-up after month 12, there were minimal changes in FFMI regardless of the initial fat-free mass level at diagnosis.
However, the slopes were significantly different during the first three months between
317
patients that presented with reduced fat-free mass and those that had normal fat-free mass at diagnosis for the overall population model and in stratified models according to HIV status (Tables 10:6 and 10:7).
Fat mass
When HIV was replaced by gender among the adjusters in multivariable mixed effects two spline model at month 3 and at month 12 with FMI as the dependent variable, the total population average for FMI at baseline, month 3, and month 12, was -2.69 (95% CI:
-3.04, -2.33), -2.03 (95% CI: -2.45, -1.62), and -2.26 (95% CI: -2.74, -1.77) significant lower among patients that presented with reduced fat mass compared to those that had normal fat mass at diagnosis, respectively. On average in stratified models among HIV positive individuals at baseline, month 3, and month 12, FMI was -2.59 (95% CI: -3.06, -
2.12), -2.18 (95% CI: -2.73, -1.63), and -2.28 (95% CI: -2.09, -1.60) significant lower among individuals that presented with reduced fat mass compared to those that had normal fat mass at diagnosis, respectively. Similarly, among HIV negative indivdiuals at baseline, month 3, and month 12, FMI was -2.85 (95% CI; -3.40, -2.30), -1.77 (95% CI: -
2.36, -1.17), and -2.23 (95% CI: -2.90, -1.56) for individuals that presented with reduced
fat mass compared to patients that had normal fat mass, respectively.
During the first three months in the overall two spline model, patients that presented with
reduced fat mass had a significantly gain in FMI at rate of 2.10 (95% CI: 0.25, 3.96)
kg/m2 per month whereas patients that presented with normal fat mass had minimal
change in FMI at a rate of -0.53 (95% CI: -2.32, 1.27) kg/m2 per month (Tables 10:6 and
318
10:7). In stratified mixed effects two spline models according to HIV status, there were
minimal changes in FMI regardless of the initial fat mass level. However, the slopes were
significantly different during the first three months between patients that presented with
reduced fat mass and those that had normal fat mass at diagnosis for the overall
population model and in stratified models according to HIV status. After month 3 and
before month 12 in the overall population model and in stratified models according to
HIV status, patients that presented with normal fat mass at diagnosis, had significant gain
in FMI; however, there were no differences in slopes between the rate of increase in
patients that presented with reduced fat mass and those that had normal fat mass (Tables
10:6 and 10:7).
Body mass index
The population average BMI at baseline, month 3, and month 12 in multivariable mixed
effects two spline model at month 3 and at month 12 with BMI as the dependent variable
when HIV was replaced by gender among the adjusters was -3.59 (95% CI; -3.98, -3.20),
-3.22 (95% CI; -3.63, -2.80), and -3.47 (95% CI; -4.02, -2.03) kg/m2 significantly lower
among patients that presented with reduced BMI compared to patients that had normal
BMI, respectively. On average among HIV negative individuals in stratified models at
baseline, month 3, and month 12, BMI was -3.49 (95% CI: -4.08, -2.91), -2.91 (95% CI: -
3.56, -2.26), and -3.15 (95% CI: -3.95, -2.34) significant lower among individuals that presented with reduced BMI compared to those that had normal BMI at diagnosis, respectively. Similarly, among HIV positive individuals at baseline, month 3, and month
319
12, BMI was -3.70 (95% CI; -4.23, -3.17), -3.48 (95% CI: -4.01, -2.95), and -3.70 (95%
CI: -4.43, 2.97) for individuals that presented with reduced BMI compared to individuals
that had normal BMI, respectively.
In the overall two spline model during the first three months, patients that presented with
reduced BMI had a significantly gain in BMI at rate of 2.87 (95% CI: 0.78, 4.96) kg/m2
per month whereas patients that presented with normal BMI had minimal gain in BMI at
a rate of 1.38 (95% CI: -0.42, 3.17) kg/m2 per month (Tables 10:6 and 10:7). In stratified mixed effects two spline models, the gain in BMI among patients that presented with reduced BMI at diagnosis was however substantial in HIV positive individuals at rate of
3.16 (95% CI: 0.51, 5.82). The gain was minimal among HIV negative individuals that presented with reduced BMI with a rate of 3.39 (95% CI: -0.03, 6.80). The slopes however were not different during the first three months between patients that presented with reduced BMI and those that had normal BMI at diagnosis for HIV positive individuals in the stratified model. There were marginal changes in BMI after month 3 and during the one year of follow-up after month 12 in the total population and in stratified models for HIV negative and HIV negative individuals (Tables 10:6 and 10:7).
Discussion
To our knowledge, this is the first study to report a two year follow-up of changes in body composition and how the changes differ by gender and by HIV status among tuberculosis patients in urban Uganda, Kampala. In this retrospective cohort study of adult patients with pulmonary tuberculosis in urban Uganda, body wasting as measured
320
by reduced fat-free mass, fat mass, and body mass indexes was associated with a substantial linear increase in fat-free mass, fat mass and body mass index during the first three of months tuberculosis treatment, but the increase varied by gender and not by HIV status of the patient. Changes in body composition among men were affected by the initial fat-free mass and BMI whereas in women, body composition changes were affected by the initial fat mass. The body composition changes among HIV positive and
HIV negative patients with tuberculosis appear not to be influenced by the initial fat-free mass, fat mass, and BMI at diagnosis. There were minimal changes in body composition after three months of tuberculosis treatment and during the one year period of follow-up after month 12 among regardless of the initial body composition, gender and HIV status suggesting lack of catch-up to normal nutritional status for patients that present with body wasting.
Our findings appear to suggest that in the face of body wasting and malnutrition during the first three months of tuberculosis treatment, changes in body composition are affected by gender but not HIV. HIV appears to play a marginal role for changes in BMI in co- infected individuals. Men that had reduced fat-free mass and BMI at diagnosis attained dramatic linear increases in fat-free mass and BMI during the first three months of tuberculosis treatment compared to men that had normal fat-free mass and BMI.
However, women who presented with reduced fat mass had substantial linear increase in fat mass compared to women that had normal fat mass. There were no changes in FFMI
321
and FMI during the first three months of tuberculosis treatment regardless of HIV status.
There was no difference in slopes between patients that had reduced BMI and those that had normal BMI at diagnosis regardless of HIV status during the first three months of tuberculosis treatment.
This study is the first to report that the initial fat-free mass and BMI influences the subsequent changes in body composition among men with tuberculosis. However, the notion that the initial fat mass in women may influence subsequent changes in body composition has been reported previously in during HIV (Swanson et al. 2000). Results of the present study are consistent with previous studies (Paton et al. 2004) that revealed total lean tissue referred to as fat-free mass increased during the first six weeks of tuberculosis treatment and thereafter the increase was in fat mass. In our present study, there were minimal changes in body composition after month 3 with slight increase in fat mass for men and slight increase in BMI for HIV positive individuals. The prior study
(Paton et al. 2004); however, was limited by sample size of only 36 participants, the study population comprised of only patients who were wasted (BMI <18.5 kg/m2) and had no HIV, and it had a short duration of follow-up for only 6 months. The strengths of the present study include: large number of study participants comprising of women and men, both wasted and not wasted, both HIV positive and HIV negative individuals, and long duration of follow-up.
322
The present study has shown that patients who presented with reduced fat and fat-free mass gained in body composition at a higher rate than patients who presented with normal levels of fat and fat-free mass, particularly during the first three months of tuberculosis treatment. However, the gain was predominantly fat mass among women whereas men gained predominantly fat-free mass plus substantial levels of fat mass. The potential explanation appears to rest on the gender differences in fat-free mass content.
Women tend to have low fat-free mass compared to men, yet fat-free mass has been shown to be a significant determinant of fat oxidation (Nagy et al. 1996; Toth et al.
1996). In several reports after adjusting for difference in fat-free mass, resting fat oxidation is lower in women than men (Nagy et al. 1996; Horton et al. 1998; Toth et al.
1998). Thus, this gender differences in fat metabolism as explained elsewhere (Blaak
2001), is associated with lower basal fat oxidation that may contribute to the increased fat storage or gain in women compared to men. Moreover, longitudinal studies have shown that low rates of fat utilization predict subsequent weight (Zurlo et al. 1990). Besides differences in lipolysis, other factors such as differences in hormone action may contribute to differences in fat oxidation. Higher levels of androgens in men stimulate the formation of the nucleic acids essential for protein biosynthesis (Mooradian, Morley, and
Korenman 1987) with eventual increase in lean tissue.
In normal healthy adults, protein catabolism equals protein anabolism; however, protein catabolism exceeds protein anabolism in tuberculosis (Macallan et al. 1998). In general; however, the catch-up linear increase in lean tissue during the first three months among patients that presented with reduced fat-free mass levels suggest that patients with
323
tuberculosis can mount a protein anabolic response during treatment (Paton et al. 2003;
Macallan et al. 1998). It may be possible that the degree of body wasting offsets the
stimulus of tuberculosis on protein metabolism till a net state of anabolism (Tomkins et
al. 1983) even as the patient become sterile from the causative organism during
treatment. In the present study, there minimal changes in fat-free mass regardless of
gender and HIV status after month 3 on follow-up. Alternatively, the response could be explained by the effective adaptive response of protein metabolism to chronic inflammatory state (Paton et al. 2001).
Despite the dramatic linear increase in fat-free mass, fat mass, and BMI among patients who presented with body wasting, these patients did not normalize their indices for body composition. That is they did not regain body composition comparable to patients presenting without wasting. There may be several explanations for this. It is possible that patients who present with wasting at the time of tuberculosis diagnosis were of slim body build before disease as compared to the patients that present with normal nutritional status. While on tuberculosis treatment, the patients with apparent regain the original body composition status prior to onset of tuberculosis disease. Following attainment of the original body composition, there are minimal changes in lean tissue and individuals adapt effective energy-sparing mechanism in balance with the usual energy intake
(Kurpad, Muthayya, and Vaz 2005; Ferro-Luzzi et al. 1997). It is also possible that households where tuberculosis occurs in Uganda, and in other countries in sub-Saharan region, there is food insecurity compromising energy intake. Thus, persistently low body composition may mark lack of access to sufficient calorie and protein intake. The
324
persistently low body composition measures even after effective tuberculosis treatment
may be a marker for future health risks. Among patients with wasting who do not
normalize, there may be consequences regarding survival and physical function in the
event of any future disease insult. Further research is needed to understand the health risk
among these patients.
In this study, we used the BIA method in measurement of body composition, yet it is not
of reference standard like the dual-energy x-ray absorptiometry. The BIA prediction method used has not yet been validated in the local population. As a result, findings of body composition may be biased because of variations in hydration across ethnic groups
(Kyle et al. 2004). However, the equations that were used in this study were previously cross validated in individuals of different race (white, black, and Hispanic) among men and women, who were both healthy controls and HIV-infected patients (Kotler et al.
1996). Moreover, the equations have been used widely in other studies from Africa with meaningful findings (Shah et al. 2001; Van Lettow et al. 2004; Villamor et al. 2006;
Mupere et al. 2010). We also took care to take measurements at rest, with proper placement of leads, in participants who had not exercised or taken alcohol, in participants with voided bladder and ambient temperature. However, measurements were in patients with underlying illness that may cause shifts in body water compartments, thereby affecting measurements of fat mass. Our findings also limited by the lack of dietary intake assessment to give further insight in the interpretation of gender differences in longitudinal body composition changes.
325
Despite limitations of the present study, findings in this study revealed remarkable gender
but not HIV differences in longitudinal body composition changes during the initial phase
of treatment among tuberculosis patients that presented with body wasting and
malnutrition. Body composition changes among men were affected by the initial fat-free
mass whereas among women by fat mass. Further evaluation is needed to under the impact of providing nutritional interventions as adjuvant treatment on body composition among TB patients in sub-Saharan and evaluation of nutritional status should be involve
BMI, fat and fat-free mass.
Acknowledgements
We thank all study staff members of the Case Western Reserve University and Makerere
University research collaboration at the Tuberculosis Research Unit in the United States
and in Uganda for their assistance; the faculty of staff at Case Western Reserve
University Department of Epidemiology and Biostatistics for the guidance in analyzing
the project; and the Fogarty International Center, for the continued support.
This study was supported in part by the AIDS International Training Research Program,
Fogarty International Center, Grant No. TW000011, based at Case Western Reserve
University, Department of Epidemiology and Biostatistics, and Tuberculosis Research
326
Unit (grant N01-AI95383 and HHSN266200700022C/ N01-AI70022 from the NIAID).
This work was part of Ezekiel Mupere’s PhD thesis at Case Western Reserve University.
327
Table 10:2 Baseline characteristics of the study population with/without baseline wasting
FFMI FMI BMI
Characteristic Normal Wasting Normal Wasting Normal Wasting
(n=184) (n=94) (n=158) (n=120) (n=424) (n=293)
n (%) n (%) n (%) n (%) n (%) n (%)
Age (years)
≤30 122 (66) 54 (57) 99 (63) 77 (64) 252 (59) 170 (58)
>30 62 (34) 40 (43) 59 (37) 43 (36) 172 (41) 123 (42)
Sex
Female 115 (62) 19 (20)a 65 (41) 69 (58)b 232 (55) 115 (39)a
Male 69 (38) 75 (80) 93 (59) 51 (42) 192 (45) 178 (61)
HIV-serostatus
Negative 105 (57) 53 (56) 90 (57) 68 (57) 198 (47) 140 (48)
Positive 79 (43) 41 (44) 68 (43) 52 (43) 226 (53) 153 (52)
HGB (mg/dl)1
>10 140 (76) 72 (77) 135 (85) 77 (64)a 208 (81) 118 (64)a
328
≤10 44 (24) 22 (23) 23 (15) 43 (36) 48 (19) 67 (36)
Smoker2
No 159 (87) 56 (60)a 127 (80) 88 (74) 364 (86) 210 (72)a
Yes 24 (13) 38 (40) 31 (20) 31 (26) 58 (14) 82 (28)
Takes alcohol3
No 135 (74) 59 (63) 106 (67) 88 (74) 259 (61) 192 (66)
Yes 48 (26) 35 (37) 52 (33) 31 (26) 164 (39) 101 (34)
Extent CXR4
38 (21) 8 (9)b 26 (17) 20 (17) 74 (18) 33 (11)b Normal/minimal
Mod/advanced 143 (79) 86 (91) 131 (83) 98 (83) 346 (82) 257 (89)
Weight loss5
No 32 (17) 12 (13) 31 (20) 13 (11) 107 (25) 52 (18)b
Yes 151 (83) 82 (87) 126 (80) 107 (89) 315 (75) 240 (82) ap-value <0.001, bp-value <0.05. BMI = body mass index, FFMI = fat-free mass index, FMI = fat mass index. 1276 missed hemoglobin measurement due to lack of blood for BMI data; 4three missed extent variable in BIA data and 7 in BMI data; 3one missed history of prior smoking in BIA data and 3 missed in BMI data; 3one missed history of alcohol intake for BMI and BIA data; 5one missed history of weight loss for BIA data and 3 missed in BMI data; FFMI wasting (<16.7 kg/m2 for men and <14.6 kg/m2 for women, normal (≥ 16.7kg/m2 for men, ≥14.6 kg/m2 for women); reduced BMI <18.5 kg/m2.
329
Table 10:3 Unadjusted rate of change for fat-free mass (FFMI), fat mass (FMI), and body mass index (BMI) among pulmonary patients with reduced FFMI, FMI, and
BMI in Kampala, Uganda
Characteristic Rate SE 95% CI
Baseline fat-free mass index
Not wasted, slope before 3 mo 1.59 0.32 0.96, 2.23
Wasted, slope before 3 mo 4.02 0.45 3.14, 4.91
Not wasted, slope after 3 mo 0.06 0.11 -0.15, 0.28
Wasted, slope after 3 mo -0.07 0.16 -0.39, 0.24
Not wasted, slope after 12 mo 0.11 0.09 -0.06, 0.29
Wasted, slope after 12 mo 0.06 0.14 -0.21, 0.33
Difference in slope before 3 mo 2.43 0.56 1.34, 3.52
Difference in slope after 3 mo -0.14 0.19 -0.52, 0.24
Difference in slope after 12 mo -0.05 0.16 -0.37, 0.27
Baseline fat mass index
Not wasted, slope before 3 mo 0.58 0.36 -0.12, 1.29
Wasted, slope before 3 mo 2.99 0.41 2.18, 3.80
330
Not wasted, slope after 3 mo 1.07 0.16 0.77, 1.38
Wasted, slope after 3 mo 1.02 0.18 0.67, 1.38
Not wasted, slope after 12 mo 0.23 0.13 -0.03, 0.49
Wasted, slope after 12 mo 0.39 0.15 0.10, 0.68
Difference in slope before 3 mo 2.41 0.55 1.33, 3.48
Difference in slope after 3 mo -0.05 0.24 -0.52, 0.42
Difference in slope after 12 mo 0.17 0.20 -0.22, 0.55
Baseline body mass index
Not wasted, slope before 3 mo 3.94 0.32 3.32, 4.56
Wasted, slope before 3 mo 6.66 0.38 5.91, 7.41
Not wasted, slope after 3 mo 0.94 0.13 0.69, 1.18
Wasted, slope after 3 mo 0.72 0.15 0.42, 1.02
Not wasted, slope after 12 mo 0.30 0.10 0.10, 0.50
Wasted, slope after 12 mo 0.20 0.12 -0.05, 0.40
Difference in slope before 3 mo 2.72 0.50 1.74, 3.69
Difference in slope after 3 mo -0.22 0.20 -0.61, 0.17
Difference in slope after 12 mo -0.11 0.16 -0.42, 0.21
331
Fat-free mass wasting = FFMI <16.7 kg/m2 for men and <14.6 kg/m2 for women, fat mass wasting = FMI <1.8 kg/m2 for men and <3.9 kg/m2 for women; reduced BMI <18.5 kg/m2 for women and men.
332
Table 10:4 Adjusted rate of change for fat-free mass (FFMI), fat mass (FMI), and body mass index (BMI) among tuberculosis patients presenting with reduced FFMI,
FMI, and BMI in Kampala, Uganda
Characteristics Overall model
Rate SE 95% CI
Baseline fat-free mass index
Not wasted, slope before 3 mo 0.35 0.88 -1.38, 2.08
Wasted, slope before 3 mo 2.68 1.05 0.68, 4.67
Not wasted, slope after 3 mo 0.52 0.31 -0.09, 1.13
Wasted, slope after 3 mo 0.41 0.36 -0.30, 1.12
Not wasted, slope after 12 mo 0.02 0.23 -0.44, 0.48
Wasted, slope after 12 mo -0.05 0.28 -0.60, 0.50
Difference in slope before 3 mo 2.33 0.57 1.20, 3.45
Difference in slope after 3 mo -0.10 0.20 -0.51, 0.30
Difference in slope after 12 mo -0.07 0.17 -0.40, 0.26
Baseline fat mass index
333
Not wasted, slope before 3 mo -0.11 0.93 -1.94, 1.73
Wasted, slope before 3 mo 2.23 0.98 0.30, 4.16
Not wasted, slope after 3 mo 1.18 0.41 0.38, 1.98
Wasted, slope after 3 mo 1.08 0.43 0.23, 1.93
Not wasted, slope after 12 mo 0.38 0.31 -0.24, 0.99
Wasted, slope after 12 mo 0.58 0.34 -0.08, 1.24
Difference in slope before 3 mo 2.33 0.55 1.26, 3.41
Difference in slope after 3 mo -0.09 0.24 -0.57, 0.38
Difference in slope after 12 mo 0.20 0.20 -0.19, 0.59
Baseline body mass index
Not wasted, slope before 3 mo 2.25 0.98 0.33, 4.17
Wasted, slope before 3 mo 3.83 1.10 1.67, 5.99
Not wasted, slope after 3 mo 1.22 0.44 0.36, 2.07
Wasted, slope after 3 mo 0.79 0.49 -0.17, 1.76
Not wasted, slope after 12 mo 0.23 0.35 -0.46, 0.91
Wasted, slope after 12 mo 0.08 0.40 -0.71, 0.87
Difference in slope before 3 mo 1.58 0.60 0.40, 2.75
334
Difference in slope after 3 mo -0.43 0.28 -0.97, 0.11
Difference in slope after 12 mo -0.14 0.23 -0.60, 0.32
Fat-free mass and BMI multivariable models were adjusted for HIV, status of anemia, prior smoking status, history of weight loss, and extent of disease on chest x-ray. Fat mass multivariable model was adjusted for HIV, prior smoking status, history of weight loss, and extent of disease on chest x-ray. Fat-free mass wasting = FFMI <16.7 kg/m2 for men and <14.6 kg/m2 for women, fat mass wasting = FMI <1.8 kg/m2 for men and <3.9 kg/m2 for women; reduced BMI <18.5 kg/m2 for women and men.
335
Table 10:5 Adjusted rate of change for fat-free mass (FFMI), fat mass (FMI), and body mass index (BMI) among tuberculosis patients presenting with reduced FFMI,
FMI, and BMI stratified according to gender in Kampala, Uganda
Stratified models
Characteristics Women Men
Rate SE 95% CI Rate SE 95% CI
Baseline fat-free mass index
Not wasted, slope before 3 mo -0.21 1.03 -2.24, 1.81 2.67 1.62 -0.52, 5.85
Wasted, slope before 3 mo 2.07 1.43 -0.74, 4.88 4.55 1.67 1.26, 7.83
Not wasted, slope after 3 mo 0.49 0.38 -0.25, 2.23 0.57 0.54 -0.50, 1.63
Wasted, slope after 3 mo 0.50 0.55 -0.57, 1.58 0.28 0.57 -0.84, 1.40
Not wasted, slope after 12 mo 0.30 0.31 -0.32, 0.92 -0.25 0.37 -0.98, 0.47
Wasted, slope after 12 mo 0.13 0.49 -0.83, 1.09 -0.28 0.39 -1.05, 0.50
Difference, slope before 3 mo 2.29 0.97 0.38, 4.19 1.88 0.77 0.36, 3.39
Difference, slope after 3 mo 0.01 0.39 -0.75, 0.77 -0.29 0.26 -0.79, 0.22
Difference, slope after 12 mo -0.17 0.36 -0.87, 0.53 -0.02 0.19 -0.40, 0.35
Baseline fat mass index
336
Not wasted, slope before 3 mo 0.13 1.38 -2.58, 2.83 1.10 1.23 -1.31, 3.51
Wasted, slope before 3 mo 3.55 1.60 0.40, 6.70 3.16 1.20 0.80, 5.52
Not wasted, slope after 3 mo 1.80 0.64 0.54, 3.07 -0.09 0.46 -1.00, 0.82
Wasted, slope after 3 mo 1.41 0.75 -0.06, 2.89 -0.24 0.46 -1.14, 0.65
Not wasted, slope after 12 mo 0.11 0.48 -0.83, 1.05 0.92 0.37 0.20, 1.65
Wasted, slope after 12 mo 0.21 0.57 -0.91, 1.33 1.03 0.35 0.35, 1.72
Difference, slope before 3 mo 3.42 0.96 1.54, 5.30 2.06 0.61 0.86, 3.25
Difference, slope after 3 mo -0.39 0.44 -1.26, 0.48 -0.15 0.23 -0.61, 0.30
Difference, slope after 12 mo 0.10 0.35 -0.59, 0.80 0.11 0.19 -0.27, 0.49
Baseline body mass index
Not wasted, slope before 3 mo 1.97 1.43 -0.83, 4.77 4.75 1.66 1.49, 8.00
Wasted, slope before 3 mo 3.30 1.74 -0.11, 6.72 6.45 1.75 3.02, 9.87
Not wasted, slope after 3 mo 1.57 0.67 0.27, 2.88 0.67 0.56 -0.42, 1.76
Wasted, slope after 3 mo 1.62 0.83 -0.00, 3.24 0.03 0.58 -1.11, 1.17
Not wasted, slope after 12 mo 0.22 0.49 -0.75, 1.18 0.51 0.48 -0.43, 1.46
Wasted, slope after 12 mo -0.11 0.62 -1.33, 1.11 0.56 0.51 -0.45, 1.57
Difference, slope before 3 mo 1.33 1.08 -0.80, 3.46 1.70 0.77 0.20, 3.20
337
Difference, slope after 3 mo 0.05 0.52 -0.98, 1.07 -0.64 0.27 -1.16, -0.12
Difference, slope after 12 mo -0.32 0.40 -1.11, 0.46 0.04 0.25 -0.44, 0.53
Fat-free mass and BMI multivariable models were adjusted for HIV, status of anemia, prior smoking status, history of weight loss, and extent of disease on chest x-ray. Fat mass multivariable model was adjusted for HIV, prior smoking status, history of weight loss, and extent of disease on chest x-ray. Fat-free mass wasting = FFMI <16.7 kg/m2 for men and <14.6 kg/m2 for women, fat mass wasting = FMI <1.8 kg/m2 for men and <3.9 kg/m2 for women; reduced BMI <18.5 kg/m2 for women and men.
338
Table 10:6 Adjusted rate of change for fat-free mass (FFMI), fat mass (FMI), and body mass index (BMI) among tuberculosis patients presenting with reduced FFMI,
FMI, and BMI in Kampala, Uganda
Characteristics Overall model
Rate SE 95% CI
Baseline fat-free mass index
Not wasted, slope before 3 mo -0.35 0.87 -2.05, 1.36
Wasted, slope before 3 mo 1.59 1.04 -0.46, 3.64
Not wasted, slope after 3 mo 0.34 0.31 -0.27, 0.95
Wasted, slope after 3 mo 0.15 0.38 -0.60, 0.90
Not wasted, slope after 12 mo 0.11 0.23 -0.34, 0.57
Wasted, slope after 12 mo 0.06 0.30 -0.53, 0.65
Difference in slope before 3 mo 1.93 0.60 0.75, 3.11
Difference in slope after 3 mo -0.19 0.22 -0.62, 0.23
Difference in slope after 12 mo -0.05 0.18 -0.40, 0.30
Baseline fat mass index
339
Not wasted, slope before 3 mo -0.53 0.91 -2.32, 1.27
Wasted, slope before 3 mo 2.10 0.94 0.25, 3.96
Not wasted, slope after 3 mo 1.60 0.39 0.82, 2.37
Wasted, slope after 3 mo 1.30 0.41 0.49, 2.10
Not wasted, slope after 12 mo 0.40 0.31 -0.20, 1.00
Wasted, slope after 12 mo 0.59 0.32 -0.05, 1.23
Difference in slope before 3 mo 2.63 0.56 1.53, 3.73
Difference in slope after 3 mo -0.30 0.24 -0.77, 0.17
Difference in slope after 12 mo 0.19 0.20 -0.20, 0.58
Baseline body mass index
Not wasted, slope before 3 mo 1.38 0.91 -0.42, 3.17
Wasted, slope before 3 mo 2.87 1.07 0.78, 4.96
Not wasted, slope after 3 mo 1.35 0.41 0.55, 2.16
Wasted, slope after 3 mo 1.01 0.48 0.07, 1.96
Not wasted, slope after 12 mo 0.14 0.33 -0.51, 0.78
Wasted, slope after 12 mo 0.01 0.40 -0.76, 0.79
Difference in slope before 3 mo 1.49 0.61 0.31, 2.68
340
Difference in slope after 3 mo -0.34 0.28 -0.88, 0.20
Difference in slope after 12 mo -0.12 0.24 -0.59, 0.34
Adjusted for HIV,age, status of anemia, smoking status, history of weight loss, and extent of disease on chest x-ray
341
Table 10:7 Adjusted rate of change for fat-free mass (FFMI), fat mass (FMI), and body mass index (BMI) among tuberculosis patients presenting with reduced FFMI,
FMI, and BMI stratified according to HIV in Kampala, Uganda
Stratified models
Characteristics HIV negative HIV positive
Rate SE 95% CI Rate SE 95% CI
Baseline fat-free mass index
Not wasted, slope before 3 mo -0.26 1.04 -2.31, 1.79 -0.58 1.58 -3.69, 2.53
Wasted, slope before 3 mo 1.40 1.26 -1.08, 3.88 1.84 1.89 -1.88, 5.56
Not wasted, slope after 3 mo 0.02 0.40 -0.77, 0.81 0.71 0.54 -0.35, 1.77
Wasted, slope after 3 mo -0.20 0.49 -1.16, 0.76 0.55 0.65 -0.73, 1.83
Not wasted, slope after 12 mo 0.26 0.23 -0.18, 0.71 -0.19 0.48 -1.15, 0.76
Wasted, slope after 12 mo 0.19 0.28 -0.37, 0.75 -0.25 0.66 -1.55, 1.04
Difference, slope before 3 mo 1.66 0.72 0.24, 3.07 2.42 0.99 0.46, 4.38
Difference, slope after 3 mo -0.22 0.27 -0.76, 0.32 -0.16 0.34 -0.83, 0.51
Difference, slope after 12 mo -0.07 0.16 -0.39, 0.24 -0.06 0.39 -0.82, 0.71
Baseline fat mass index
342
Not wasted, slope before 3 mo 0.33 0.98 -1.60, 2.27 -2.54 1.76 -6.00, 0.93
Wasted, slope before 3 mo 1.97 1.04 -0.07, 4.01 1.79 1.80 -1.76, 5.35
Not wasted, slope after 3 mo 1.26 0.55 0.18, 2.34 2.04 0.64 0.78, 3.30
Wasted, slope after 3 mo 1.13 0.58 -0.06, 2.27 1.42 0.65 0.14, 2.70
Not wasted, slope after 12 mo 0.49 0.41 -0.31, 1.29 -0.15 0.49 -1.11, 0.81
Wasted, slope after 12 mo 0.41 0.42 -0.42, 1.24 0.56 0.51 -0.46, 1.57
Difference, slope before 3 mo 1.63 0.57 0.51, 2.76 4.33 1.03 2.30, 6.36
Difference, slope after 3 mo -0.13 0.31 -0.73, 0.48 -0.62 0.38 -1.37, 0.13
Difference, slope after 12 mo -0.08 0.23 -0.54, 0.37 0.71 0.34 0.03, 1.38
Baseline body mass index
Not wasted, slope before 3 mo 1.06 1.51 -1.90, 4.02 2.26 1.14 0.02, 4.51
Wasted, slope before 3 mo 3.39 1.74 -0.03, 6.80 3.16 1.35 0.51, 5.82
Not wasted, slope after 3 mo 1.06 0.63 -0.18, 2.90 1.68 0.52 0.67, 2.70
Wasted, slope after 3 mo 0.74 0.72 -0.68, 2.16 1.38 0.63 0.15, 2.62
Not wasted, slope after 12 mo 0.20 0.46 -0.70, 1.10 0.15 0.43 -0.69, 0.99
Wasted, slope after 12 mo 0.11 0.54 -0.94, 1.17 0.21 0.54 -0.85, 1.28
Difference, slope before 3 mo 2.33 0.96 0.45, 4.20 0.90 0.78 -0.63, 2.43
343
Difference, slope after 3 mo -0.32 0.40 -1.09, 0.46 -0.30 0.37 -1.03, 0.43
Difference, slope after 12 mo -0.09 0.30 -0.68, 0.51 0.06 0.34 -0.60, 0.73
Adjusted for HIV,age, status of anemia, smoking status, history of weight loss, and extent of disease on chest x-ray
344
Figure 10:1 Overall Study flow diagram
Total 745 assembled from: HHC 314 +
28 Excluded: <18 years 717 Involved in analysis
Normal BMI 424 (59%) Reduced BMI 293 (41%)
HIV positive HIV positive 226 (53%) 153 (52%) HIV negative HIV negative 198 (47%) 140 (48%) Women Women 232 (55%) 115 (39%) Men 192 (45%) Men 178 (61%)
Repeated BMI measurements at: Baseline, 2, 3, 5, 6, 12, and 24 month Total observation = 4,875 Missing observation = 781
Reduced BMI <18.5 kg/m2for men and women.
345
Figure 10:2 Study flow diagram for BIA data
341 KCH with BIA data subset of 745
63 Excluded: <18 years 278 Involved in analysis
Normal FFMI 184 (66%) Reduced FFMI94 (34%)
HIV positive 79 (43%) HIV positive 41 (44%) HIV negative 105 (57%) HIV negative 53 (56%) Women 115 (62%) Women 19 (20%)
Men 69 (38%) Men 75 (80%)
Normal FMI 158 (57%) Reduced FMI 120 (43%)
HIV positive 68 (43%) HIV positive 52 (43%) HIV negative 90 (57%) Women 65 (41%) HIV negative 68 (57%) Men 93 (59%) Women 69 (58%)
Repeated FFMI or FMI measurements at: Baseline, 3, 12, and 24 month; Total observation = 1112 Missing observation = 205
FFMI wasting (<16.7 kg/m2 for men and <14.6 kg/m2 for women, normal (≥ 16.7kg/m2 for men, ≥14.6 kg/m2 for women).
346
CHAPTER 11
SUMMARY OF RESULTS AND GENERAL CONCLUSIONS
347
DISCUSSION
Introduction
Body wasting is a cardinal feature of tuberculosis and is probably one of the determinants of disease severity and outcome. However, its etiology and its management are poorly
understood, and its assessment is overlooked in research and in clinical practice. Body
wasting is loss of body mass, most notably muscle or lean body mass referred to as fat-
free mass. The simple and portable bioelectrical impedance analysis (BIA) device
provides the precise and practical approach of partitioning body weight into fat and fat-
free mass in clinical medicine. However, it may not be easily affordable and accessible in
sub-Saharan Africa where it is needed to detect body wasting early enough among tuberculosis and HIV patients. Sub-Saharan Africa has a high burden of tuberculosis and
HIV-associated tuberculosis. The widely used body mass index (BMI) in assessing body wasting is popularly a measure of body fat and it gives unreliable results of wasting when
it is low. Thus, BMI overestimates body wasting. BMI can be partitioned into body fat
mass (FMI) and fat-free mass (FFMI) indices by dividing fat and fat-free mass with
height. Body composition in this dissertation refers to fat and fat-free mass compartments
of the body weight.
The data that was used to generate results in this dissertation were obtained from cross-
sectional and retrospective cohort designs. The cross-sectional design enrolled 131 HIV positive and HIV negative adults with or without tuberculosis in urban Uganda. In this
348
design, participants were assessed for waist circumference (WC), BMI, mid-upper-arm
circumference (MUAC), and the 24-hour dietary intake recall. In a retrospective cohort
design, 745 adult tuberculosis patients that were merged from three completed NIH
funded prospective studies at the Uganda-Case Western Reserve University research collaboration were analyzed. All patients were screened for HIV infection and had BMI
measurements at month 0, 2, 3, 5, 6, 12, and 24 on follow-up. A subset (314) of the 745
had BIA fat and fat-free mass measurements at month 0, 3, 12, and 24 on follow-up. Any case of deaths and date was documented on follow-up.
Purpose
The purpose of this dissertation was to generate information that could be used to improve survival in tuberculosis patients. This dissertation focused first on understanding simple and inexpensive approaches of assessing body wasting. Second, focused on generating information that could aid in understanding of body wasting and its management, that is whether nutritional factors influence body composition in the face of tuberculosis; and whether body wasting as measured by precise measures of body composition modifies the course of tuberculosis.
349
SUMMARY OF RESULTS
Chapter Four: Body Composition Measured with Bioelectrical Impedance Analysis and Anthropometry among HIV Positive and HV Negative Adults with or without
Tuberculosis in Urban Kampala, Uganda
Analysis in this chapter established whether known existing equations that estimated body composition among Caucasians could provide comparable results of body composition to those of BIA device in an African population in Uganda. The equations involved simple measurements such as WC, BMI, and MUAC. Findings among 131 HIV positive and HIV negative adults with or with tuberculosis revealed that existing equations that provided comparable results of body composition to those of BIA differed by gender and tuberculosis status. The equation that involved MUAC or BMI provided comparable results of lean tissue to that of BIA whereas the equation that involved
MUAC provided comparable results of fat mass among men and women with tuberculosis regardless of HIV status. Among men without tuberculosis, the equation that involved MUAC provided comparable results of fat and lean tissue to that of BIA whereas among women without tuberculosis, the equation that involved WC provided comparable results of fat and lean tissue regardless of HIV status.
350
Chapter Five: Indicators of Dietary Adequacy among HIV Positive and HIV
Negative Adults in Urban Kampala, Uganda
Analysis in this chapter involved 131 HIV positive and HIV negative adults with or
without tuberculosis from urban Uganda to assess nutritional adequacy of dietary intake and the validity of simple low-cost methods to evaluate nutritional adequacy of diets consumed among HIV positive and HIV negative adults with or without tuberculosis in urban Uganda. Findings revealed that all participants (100%) consumed at least cereals,
roots, and tubers, and 90% consumed vegetables not rich in vitamin A such tomatoes and
onions while only 45% consumed vitamin-A-rich fruits and vegetables, and only 15%
consumed eggs.
The mean FVS and DDS for the study population were low 8.1 ± 2.8 and 4.7 ± 1.4,
respectively. Both men and women regardless of tuberculosis and HIV status, had
carbohydrate and ascorbic acid deficiency in the range of 0 to 30% whereas other nutrient
intakes including energy, protein, dietary fiber, calcium, magnesium, zinc, iron, vitamin
A, vitamin D, and folate had deficiencies ranging 25% to 100%. When a MAR of 65%
was used as a cut-off point for nutrient adequacy, it was found that FVS must be 9 or
more and DDS must at least 5. Among women, both FVS and DDS had a high ability to
identify participants with an inadequate or adequate diet while among men FVS had a
high ability to identify individuals with inadequate diet but low ability to identify those
with adequate diet, DDS had low ability to identify individuals with inadequate diet but
had a high ability to identify those with adequate diet.
351
Chapter Six: Predictors of Fat and Lean Tissue among HIV Positive and HIV
Negative Adults with or without Tuberculosis in Urban Kampala, Uganda.
Analysis in this chapter aimed to establish factors that may influence fat mass, fat-free
mass (lean tissue), and BMI among 131 HIV positive and HIV negative adults with or
without tuberculosis in urban Uganda. We found that energy intake was associated with
an increase in BMI among women (0.003 ± 0.001 SE, p=0.028) although intake in the
presence of tuberculosis was associated with a decrease in BMI (-0.004 ± 0.001 SE,
p=0.008). Protein intake among women with no income (-0.02 ± 0.01 SE, p=0.027) and
among unemployed women (-0.08 ± 0.03 SE, p=0.010) was associated with a decrease in
lean tissue and fat mass, respectively whereas protein intake among women with
tuberculosis was associated with an increase in BMI (0.10 ± 0.05 SE, p=0.039). Being a
single woman (0.58 ± 0.25 SE, p=0.022) was associated with an increase in lean tissue
whereas having reduced appetite (-0.79 ± 0.33 SE, p=0.020) was associated with a
decrease in lean tissue and fat mass (-4.17 ± 1.94 SE, p=0.036). Among men, tuberculosis
(-1.42 ± 0.45 SE, p=0.003) was associated with a decrease in lean tissue. Similarly, having reduced appetite was associated with decrease in fat mass (-1.90 ± 0.90 SE, p=0.040) and BMI (-2.95 ± 0.68 SE, p<0.001). HIV did not influence body composition regardless of gender.
352
Chapter Seven: Body Wasting and Dietary Intake among HIV Positive and HIV
Negative Adults with or without Tuberculosis in Urban Kampala, Uganda
The analysis in this chapter involved 131 HIV positive and HIV negative adults with or with tuberculosis to evaluate the independent effects of tuberculosis and HIV, to evaluate whether dietary intake differs by body wasting and severity of tuberculosis disease. We found that tuberculosis patients that had moderate/or severe clinical disease had lower dietary intake for energy, protein, total fat, carbohydrate, calcium, vitamin A, and folate compared to patients with mild disease. Both men and women had comparable dietary intake among patients with TB regardless of HIV status whereas HIV negative women had reduced energy, protein, and folate intake among individuals without TB compared to men. Tuberculosis patients with wasting of lean tissue or those with reduced BMI had comparable nutrient intake with counterparts that had normal lean tissue or normal BMI.
Chapter Eight: Correlates of Dietary Adequacy among HIV Positive and HIV
Negative Adults with or without Tuberculosis in Urban Kampala, Uganda
In this chapter, we evaluated dietary correlates of energy and protein intake and correlates of inadequate nutrient intake. There was female gender interaction between having tuberculosis and reduced appetite, and between having tuberculosis and being a current alcohol taker in the model for energy intake. Women that had tuberculosis with reduced appetite or tuberculosis with history of taking alcohol had decreased energy intake. Also women who had history of alcohol intake had decreased protein intake. There was no compromise with energy and protein intake among men. Women were associated with
353
inadequate iron intake. Further, women with tuberculosis were associated with inadequate folate intake. Individuals with tuberculosis residing in households of more than two people or those with no or low education were associated with inadequate vitamin A intake.
Chapter Nine: Impact of Body Wasting on Survival among Adult Patients with
Pulmonary Tuberculosis in Urban Kampala, Uganda
In this chapter, we evaluated the impact of HIV and body wasting on survival in
tuberculosis patients using precise measures of nutritional status, the height-normalized
fat-free mass (FFMI) and fat mass (FMI) indices. During the follow-up period, 19% of
310 patients with baseline wasting by BMI died compared to 11% of 437 without
wasting, a crude risk ratio of 1.74 (95% confidence interval (CI): 1.22, 2.48). Of 103 with
baseline wasting by fat-free mass index, 16% died, compared to 7% without wasting,
crude risk ratio of 2.31 (95% CI: 1.10, 3.92).
In stratified survival analysis, survival proportion was significantly lower among men
with reduced BMI compared to men with normal BMI; and lower among women with
reduced fat-free mass index compared to women with normal fat-free mass index. In
multivariable Cox regression model using anthropometric data, the relative hazard of
death when patient had reduced BMI was 1.85 (95% CI: 1.25, 2.73). In a nested model,
the relative hazard for death was 1.70 (95% CI: 1.03, 2.81) for men with reduced BMI
and 1.83 (95% CI: 0.96, 3.50) for women with reduced BMI. In a model using fat-free
354
mass index data, the relative hazard of death when patients had reduced fat-free mass index was 1.88 (0.96, 3.65). In a nested model, the relative hazard of death was 6.83
(95% CI: 2.14, 21.74) for women with reduced fat-free mass index compared to women with normal fat-free mass and 0.80 (95% CI: 0.35, 1.84) for men with reduced fat-free mass index. In Kaplan-Meier analysis, men had significantly lower survival compared to women (p=0.016) Cox regression analysis HIV positive men had 1.62 (95% confidence interval (CI): 1.05, 2.52) hazard of death compared to HIV positive women. HIV negative men had 0.57 (95% CI: 0.10, 3.31) hazard of death compared to HIV negative women.
Chapter Ten: Longitudinal Changes in Body Composition among Tuberculosis
Patients with or without Body Wasting in Urban Kampala, Uganda
In this chapter, longitudinal data analysis was conducted to assess whether body wasting as measured by height-normalized measures of nutritional status at diagnosis of tuberculosis, and whether HIV infection modifies longitudinal changes in body
composition during and after tuberculosis treatment. Results in this chapter revealed that
there were no differences in body wasting as assessed by reduced lean tissue, fat mass,
and BMI between HIV positive and HIV negative patients at diagnosis of tuberculosis. In
stratified mixed effects two spline models during the first three months of treatment, the
gain in lean tissue among patients that presented with wasted lean tissue at diagnosis was
dramatic in men with rate of 4.55 kg/m2 (95% confidence interval (CI): 1.26, 7.83) per
355
month; however, the gain was minimal among women who presented with reduced lean
tissue with rate of 2.07 kg/m2 (95% CI: -0.74, 4.88).
In stratified models for fat mass as dependent variable, women with reduced fat mass at presentation had a substantial gain in fat mass at rate of 3.55 kg/m2 (95% CI: 0.40, 6.70)
whereas men had a rate of 3.16 kg/m2 (0.80, 5.52). In stratified models with BMI as
dependent variable, men with reduced BMI at presentation gained BMI at a rate of 6.45
kg/m2 (95% CI: 3.02, 9.87) whereas women at a rate of 3.30 kg/m2 (95% CI: -0.11, 6.72).
There were minimal changes in lean tissue, fat mass, and BMI during the first three
months of treatment in stratified models according to HIV status. Further, there were
minimal changes in lean tissue, fat mass, and BMI after month 3 and during the one year
follow-up after month 12.
Conclusion
In this dissertation, we have shown that body composition can reliably be assessed using
inexpensive and easy to measure anthropometric assessments such as waist
circumference, MUAC, and BMI. Precise estimation of fat-free mass is essential to
identify patients with body wasting so that appropriate and early interventions are
instituted. The cross-sectional nature of the study limits making conclusions beyond the
time of diagnosis among tuberculosis. However, since the study population included
participants with and without tuberculosis, one can infer that the equations evaluated in
356
this dissertation can reliably be used to monitor changes in body composition during and after tuberculosis treatment.
Several conclusions were generated following our working model in this dissertation
(Figure 1). This dissertation has shown that the dietary consumption in the study population of HIV positive and HIV negative adults with or without tuberculosis from urban Uganda was monotonous, rich in carbohydrates and deficient in nutrients regardless of gender, tuberculosis, and HIV status. The ability of FVS and DDS indices to identify individuals with inadequate or adequate diet consumption differed by gender.
The FVS was a better predictor of nutritional adequacy among women whereas DDS was a better predictor among men.
Findings in this dissertation suggest that there are remarkable gender differences in how energy and protein intake influence body composition, and we found important interactions in the face of tuberculosis, and when there is no income. HIV does not appear to influence nutrient intake on body composition. At the time of tuberculosis diagnosis, the 24-hour dietary intake recall varied by severity of tuberculosis disease, but not tuberculosis disease or HIV status. However, in the absence of tuberculosis, dietary intake varied by gender. Both men and women had comparable dietary intake among patients with tuberculosis regardless of HIV status whereas HIV negative women had reduced energy, protein, and folate intake among individuals without tuberculosis compared to men. As regards dietary correlates, findings suggest that correlates of energy
357
and protein intake differ by gender. Women and individuals having tuberculosis who reside in households with two or more people or who have no or low education are at vulnerable state of inadequate nutrient intake.
We have shown in this dissertation that body wasting is associated with poor survival and this effect differed by gender. Men with reduced BMI at the time of tuberculosis had poor survival whereas women with reduced lean tissue had poor survival suggesting that BMI is a better predictor of death among men whereas wasting of lean tissue is a better predictor of death among women. Further, survival differed by gender; men had poor survival compared to women. However, this gender difference in survival was modified by HIV. It is HIV positive men that had poor survival compared to HIV positive women.
Our results suggest that body wasting at the time of tuberculosis diagnosis modifies longitudinal changes in body composition during tuberculosis treatment. The effect however, differed by gender. Men with wasted lean tissue and reduced BMI had dramatic increase in lean tissue and BMI whereas women with reduced fat mass had significant increase in fat mass during the first three months of tuberculosis treatment. HIV infection did not influence changes in body composition. Of note, there were minimal changes in body composition after month 3 and during the one year follow-up after month 12 regardless of gender, HIV status, and the initial level of body composition.
358
Figure 11:1 Overall conclusions: Tuberculosis – HIV – malnutrition model
Tuberculosis HIV Reduced appetite Immune activation Inadequate dietary intake Immune deficiency Altered metabolism: HIV replication Differ by gender Increased cytokine production Anabolic block HIV-TB Co-infection Increased energy expenditure Malabsorption Increased inadequate intake Differ by gender
Lipid and protein metabolism: Differ by gender Wasting of body composition: Differ by gender Men (Mupere et al. 2010) Women Begin with high fat-free mass Begin with low fat-free mass Begin low with fat mass Begin with high fat mass Marked fat-free mass wasting Preserve fat-free mass wasting Fat mass wasted in proportion with fat-free mass. Marked fat mass wasting Low body mass index High body mass index Gender differences in micronutrient deficiency not known?
Wasting modifies: Differ by gender Men Women Survival: Poor with low BMI Survival: Poor with low FFMI Poor HIV positive Minimal HIV positive HIV negative same as women HIV negative same as men Changes: Increase in FFM and BMI Changes: Significant increase FMI No catch-up after month 3 No catch-up after moth 3 Mechanism not known Mechanism not known Dietary intake: Differ by disease severity, not TB or HIV status or gender Predictors body composition: Energy and protein intake, TB, TB and appetite appetite, income, employment, single.
FFMI = fat-free mass index, FMI = fat mass index, BMI = body mass index.
359
Limitations and Future Studies
Findings in this dissertation, demonstrated poor survival among patients that presented with body wasting. The survival differed by gender. Men had poor survival when BMI was low, whereas women had poor survival when lean tissue was wasted. This provides compelling evidence to evaluate nutritional interventions that can improve survival among patients with body wasting. In evaluating nutritional interventions however, gender differences will need to be considered.
Our results revealed that there were minimal changes in body composition after month 3 and during the one year follow-up after month 12 regardless of gender, HIV status, or the initial level of body composition. Yet, patients that presented with wasting had dramatic increase in lean tissue and BMI for men and fat mass for women; however, there was no catch-up to normal nutritional status. Studies are needed to evaluate the mechanisms for the lack of catch-up among patients that presented with wasting and the lack of changes in body composition after month 3.
To address the limitations in this dissertation, validation studies with reference methods such as air-displacement plethysmography, underwater weighing and dual X-ray absorptiometry (DXA) are needed to validate the BIA regression equations and the existing equations that involve simple anthropometric measurements used in estimation of body composition. Validation in follow-up studies and in different populations because of variations in hydration is needed.
360
In this dissertation, a single 24-hour recall was used to assess dietary intake. However,
four multiple-pass 24-hour recalls have been shown to be the most appropriate method
for a study of diet and nutrition in low-income households (Vucic et al. 2009). A whole
year-round study is recommended to understand the habitual consumption in Uganda.
Further, evaluation is needed to understand changes in dietary intake over time and how
these changes would affect body composition and survival.
We found simple counting of food items or food groups can give fairly good assessment
of nutritional adequacy. The study however, was cross-sectional and was conducted in an
urban setting. The conclusions might therefore be valid for similar urban settings, but the
methods and approaches are equally valid for rural areas. To attain generalizability of conclusions and to understand whether dietary adequacy would change with habitual consumption, additional studies are recommended in rural settings, in different socioeconomic, and in different ethnic groups. Also follow-up studies are needed to understand how dietary adequacy changes during different seasons of the year.
Strengths
Findings in the present dissertation were generated using a large database that provided adequate power to detect differences in survival and changes in body composition by gender and body wasting. The data provided adequate power to interactions with survival between gender and HIV infection. To our knowledge, this is the first study to show the effect of wasting on survival using height-normalized fat-free mass rather than BMI that
361
may overestimate wasting among women and thus, overestimate survival due to wasting.
Height-normalized fat-free mass provide a precise measure of nutritional status than BMI
(Kyle, Piccoli, and Pichard 2003; Kyle, Genton, and Pichard 2002; VanItallie et al.
1990). The median follow-up period for participants in the data used for analysis in this
dissertation was more than two years; suggesting a longer period of follow-up than the
previous studies (Sani et al. 2006; Lucas et al. 1993). Thus, our findings are embedded with a temporal relationship between enrolment to time of death or censoring and between enrolments to different time points when changes in body composition were assessed on follow-up.
We had a full panel of control groups that enabled to this dissertation to show the
independent effects of tuberculosis and HIV on dietary intake. Further, the dissertation
has shown the biologic and socioeconomic plausibility.
Public Health Implications
Findings in this dissertation suggest the need for national tuberculosis programs to address the nutritional needs of tuberculosis patients at the time of diagnosis, during and after treatment. The nutritional needs may range from health education to provision of
nutritional supplements. Results revealed that patients with body wasting had poor
survival and those that survived after month 3, did not catch-up to normal nutritional
status. The findings provide theoretical framework for targeted nutritional intervention to
tuberculosis patients that present with body wasting, particularly women with reduced
362
lean tissue, men with reduced BMI or HIV positive men. There is potential that if
nutritional needs are addressed, this may improve survival and adherence to treatment.
Our findings contribute to an increased understanding of the potential methods of assessing body wasting using simple and inexpensive anthropometric measurements. This implies that during regular clinical practice and in public field surveys, body composition can easily be assessed to identify individuals and population groups at risk of poor outcome. Early identification of body wasting enables timely institution of management to prevent poor outcome.
The monotonous diet with substantial nutrient deficiencies in this dissertation, suggests the need for fortification of common food items in Uganda. Fortification of staple foods has been shown to improve micronutrient intake among adults in South Africa (Steyn et al. 2008). In this dissertation, we have shown that simple indices of food variety score and dietary diversity score can be used to identify individuals or families with inadequate nutrient intake. Thus, these simple scores can be used in management of tuberculosis and in public health settings as tools to provide health education so that individuals or families understand when to have an adequate diet.
To address the nutritional needs of tuberculosis patients and the nutritional needs of the community where tuberculosis patients reside, will require a collaborative effort of
363
several stakeholders including clinicians, healthcare professionals, policy makers at the
Ministry of Health in Uganda, program managers for the poverty eradication scheme and for various existing non-government organizations and the will of the political organs in the country. The clinicians, the healthcare professionals and the public health practitioners should strive to understand the reasons for poor survival, failure to attain normal nutritional status after therapy, and poor diet consumption among tuberculosis patients in order to develop and implement targeted interventions.
364
CHAPTER 12
EXTRA TABLES AND FIGURES FOR CHAPTER FOUR
365
Table 12:12:1 Mean intra-supervisor, mean intra-observer, and mean inter- observer technical error of measurement prior to data collection
Difference Technical Error of Measurement between Measurement (n=6) trainee and Intra- Intra- Inter- supervisor Supervisor Observer Observer Mean Mean Mean
Weight (kg) 0.79 0.000 0.068 0.620
Height (cm) 0.19 0.037 0.059 0.037
MUAC (cm) 0.65 0.022 0.017 0.042
Triceps ST (mm) 1.11 0.036 0.054 1.242
Biceps ST (mm) 0.39 0.029 0.049 0.155
Subscapular ST (mm) 0.30 0.034 0.057 0.088
Sacral iliac ST (mm) 0.58 0.034 0.041 0.335
Waist circumference (cm) 0.07 0.067 0.035 0.005
Hip circumference (cm) 0.43 0.043 0.026 0.184
Resistance (ohms) 0.84 0.146 0.165 0.712
Reactance (ohms) 0.23 0.044 0.046 0.051
ST = skinfold thickness, MUAC=mid-upper arm circumference.
366
Table 12:12:2 Coefficient of reliability prior to data collection
Reference(Frisancho A.R 1990;
Lohman T.G, Roche A.F, and Measurement (n=6) Martorell R 1988) Coefficient of Reliability Intra-observer Inter-observer Weight (kg) 1.200 1.500 0.99
Height (cm) 0.692 0.953 0.90
MUAC (cm) 0.347 0.425 1.00
Triceps ST (mm) 0.800 1.890 1.00
Biceps ST (mm) 0.600 0.600 0.98
Subscapular ST (mm) 1.830 1.530 0.96
Sacral iliac ST (mm) 1.000 1.700 1.00
Waist circumference (cm) 1.000 1.000 0.74
Hip circumference (cm) 1.230 1.380 1.00
Resistance (ohms) - - 0.96
Reactance (ohms) - - 0.96
ST = skinfold thickness, MUAC=mid-upper arm circumference.
367
Table 12:12:3 Guide for comparison of inter-observer error (Frisancho A.R 1990)
Measurement Difference between trainee and supervisor
Good Fair Poor Gross Error Weight (kg) 0 to 0.1 > 0.2 0.3 to 0.4 ≥ 0.5
Height (cm) 0 to 0.5 0.6 to 0.9 1.0 to 1.9 ≥ 2.0
Mid upper arm circumference (mm) 0 to 5 6 to 9 10 to 19 ≥ 20
368
Figure 12:1 Fat-free mass measured by BIA compared with fat-free mass measured by equation involving waist circumference
BIA = bioelectrical impedance analysis, FFM = fat-free mass
369
Figure 12:2 Fat mass measured by BIA compared with fat mass measured by equation involving waist circumference
BIA = bioelectrical impedance analysis, FM = fat mass, Waist = waist circumference
370
Figure 12:3 Fat-free mass measured by BIA compared with fat-free mass measured by Durnin & Womersely equations involving sum of 4 skinfold thickness
BIA = bioelectrical impedance analysis, FFM = fat-free mass, ST = skinfold thickness.
371
Figure 12:4 Fat mass measured by BIA compared with fat mass measured by
Durnin & Womersely equations that involves sum of 4 skinfold thickness
BIA = bioelectrical impedance analysis, FM = fat mass, ST = skinfold thickness.
372
Figure 12:5 Fat-free mass measured by BIA compared with fat-free mass measured by equation involving BMI
BIA = bioelectrical impedance analysis, FFM = fat-free mass, BMI = body mass index.
373
Figure 12:6 Fat mass measured by BIA compared with fat mass measured by equation involving BMI
BIA = bioelectrical impedance analysis, FM = fat mass, BMI = body mass index
374
Figure 12:7 Fat-free mass measured by BIA compared with fat-free mass measured
by equation involving MUAC
BIA = bioelectrical impedance analysis, FFM = fat-free mass, MUAC = mid-upper-arm circumference
375
Figure 12:8 Fat mass measured by BIA compared with fat mass measured by
equation involving MUAC
BIA = bioelectrical impedance analysis, FM = fat mass, MUAC=mid-upper arm circumference.
376
CHAPTER 13
EXTRA TABLES AND FIGURES FOR CHAPTER NINE
377
Table 13:1 Comparison of key baseline variables across phase II prednisolone study, household contact study, and Kawempe community health study
Household Kawempe Phase II 1Characteristics Contact study Prednisolone
[mean, (SD)] (n=95) (n=312) (n=340) p-value
Sex
Male (%) 59 (62) 163 (52) 178 (51) 0.172
Female (%) 36 (38) 151 (48) 166 (49)
HIV status
Negative (%) 0 (0) 159 (51) 186 (55) <0.001
Positive (%) 95 (100) 153 (49) 153 (45)
Age in years 30.7 (7.2) 31.8 (9.0) 29.6 (9.2) 0.003
Weight in kg 52.0 (6.8) 52.3 (8.7) 51.6 (8.4) 0.297
Height in cm 165.7 (8.8) 164.2 (9.3) 163.8 (8.7) 0.077
BMI in (kg/m2) 19.0 (2.6) 19.4 (2.9) 19.2 (2.8) 0.383
Hemoglobin in g/dl 11.0 (1.8) 11.6 (2.4) 11.4 (2.2) 0.070
Chest x-ray disease
378
Normal/mild (%) 4 (4) 54 (17) 53 (16) 0.007
Moderate/far advanced (%) 89 (96) 257 (83) 282 (84)
1Continuous variables are means ± standard deviation (SD).
379
Table 13:2 Assessing the proportional hazards assumption with a statistical test:
Correlations between ranked failure time and Schoenfeld residuals
Characteristic Mean Residuals Ranked failure time p-value
Sex -1.02x10-8 0.30 0.120
Age (years) 8.47x10-8 -0.21 0.284
HIV status 4.84x10-7 0.29 0.129
Hemoglobin (g/dl) 1.07x10-7 0.06 0.760
Fat mass index (kg/m2) -1.96x10-8 -0.05 0.788
Fat-free mass index (kg/m2) -2.16x10-8 -0.09 0.646
Body mass index (kg/m2) 2.06x10-8 -0.03 0.864
Chest x-ray disease extent -7.89x10-8 0.15 0.440
Smoker 1.75x10-8 0.06 0.766
Takes alcohol 6.80x10-8 0.24 0.215
History weight loss 5.50x10-8 -0.21 0.285
380
Table 13:3 Baseline characteristics of pulmonary tuberculosis patients between HIV negative and HIV positive
HIV negative HIV positive Characteristic (n=345) (n=401)
Sex
Females [n (%)] 165/351 (47) 186/351 (53)
Males [n (%)] 180/395 (46) 215/395 (54)
Age (years)
≤30 [n (%)] 233/434 (54) 201/434 (46)a
>30 [n (%)] 112/312 (36) 200/312 (64)
Hemoglobin (g/dl)1
>10 [n (%)] 169/342 (49) 173/342 (51)a
≤10 [n (%)] 32/121 (26) 89/121 (74)
Body mass index (kg/m2)
Normal [n (%)] 202/436 (46) 234/436 (54)
Low [n (%)] 143/310 (46) 167/310 (54)
Fat-free mass index (kg/m2)
381
Normal [n (%)] 116/207 (56) 91/207 (44)
Low [n (%)] 56/103 (54) 47/103 (46)
Fat mass index (kg/m2)
Normal [n (%)] 98/176 (56) 78/176 (44)
Low [n (%)] 71/127 (56) 56/127 (44)
Chest x-ray disease extent 46/111 (41) 65/111 (59) Normal/mild [n (%)]
Moderate/far advanced [n (%)] 296/627 (47) 331/627 (53)
Smoker
No [n (%)] 272/592 (46) 320/592 (54)
Yes [n (%)] 73/151 (48) 78/151 (52)
Currently takes alcohol
No [n (%)] 240/466 (51) 226/466 (49)a
Yes [n (%)] 105/279 (38) 174/279 (62)
History weight loss
No [n (%)] 60/165 (36) 105/165 (64)b
Yes [n (%)] 283/578 (49) 295/578 (51)
382
ap-value <0.001, bp-value <0.05.
383
Table 13:4 Multivariable Relative hazards [HR, 95% confidence intervals (CIs)] for death among tuberculosis patients with normal compared with patients having low body mass index (BMI)
Deaths/N (%) Relative 95% Confidence
Characteristic hazard Interval
Body mass index (kg/m2)
Normal 47/437 (11) 1 -
Low 58/310 (19) 13.39 1.23, 146.19
Sex
Female 38/352 (11) 1 -
Male 67/395 (17) 1.52 0.83, 2.76
Age (years)
≤30 34/434 (8) 1 -
>30 71/313 (23) 3.06 1.61, 5.83
HIV-serostatus1
Negative 6/345 (2) 1 -
Positive 99/401 (25) 42.68 5.88, 309.73
Smoker3
384
No 84/593 (14) 1 -
Yes 20/151 (13) 0.57 0.23, 1.40
Chest x-ray extent6
Normal/minimal 19/111 (17) 1 -
Moderate/far advanced 86/628 (14) 0.96 0.46, 1.99
Weight loss5
No 11/165 (7) 1 -
Yes 94/579 (16) 3.54 1.88, 6.66
Reduced BMI*sex 1.06 0.45, 2.48
Reduced BMI*HIV 0.25 0.03, 2.23
Reduced BMI*age 0.55 0.24, 1.29
Reduced BMI*smoker 1.05 0.35, 3.11
Reduced BMI*extent 0.64 0.24, 1.75
-2LL reduced model - -2LL full model: 1202.115 – 1193.495 = 8.62; df = 5, p=0.125. 1One missed HIV status; 6eight missed extent variable; 3three missed history of ever smoked; 5four missed history of weight loss; low BMI <18.5 kg/m2.
385
Table 13:5 Multivariable Relative hazards [HR, 95% confidence intervals (CIs)] for death among tuberculosis patients with normal compared with patients having low body mass index (BMI)
Deaths/N (%) Relative 95% Confidence
Characteristic hazard Interval
Body mass index (kg/m2)
Normal 47/437 (11) 1 -
Low 58/310 (19) 9.21 0.99, 85.90
Sex
Female 38/352 (11) 1 -
Male 67/395 (17) 1.52 0.85, 2.73
Age (years)
≤30 34/434 (8) 1 -
>30 71/313 (23) 3.04 1.60, 5.78
HIV-serostatus1
Negative 6/345 (2) 1 -
Positive 99/401 (25) 41.95 5.78, 304.28
Smoker3
386
No 84/593 (14) 1 -
Yes 20/151 (13) 0.60 0.36, 1.00
Chest x-ray extent6
Normal/minimal 19/111 (17) 1 -
Moderate/far advanced 86/628 (14) 0.77 0.47, 1.27
Weight loss5
No 11/165 (7) 1 -
Yes 94/579 (16) 3.53 1.88, 6.65
Reduced BMI*sex 1.03 0.45, 2.34
Reduced BMI*HIV 0.26 0.03, 2.29
Reduced BMI*age 0.56 0.24, 1.32
-2LL reduced model - -2LL full model: 1202.115 – 1194.245 = 7.87; df = 3, p=0.0488. 1One missed HIV status; 6eight missed extent variable; 3three missed history of ever smoked; 5four missed history of weight loss; low BMI <18.5 kg/m2.
387
Table 13:6 Relative hazards [HR, 95% confidence intervals (CIs)] for death among tuberculosis patients with normal compared with patients having low body mass index (BMI)
Overall model Characteristics Deaths/N (%)
HR (95% CI)
BMI (kg/m2)
Normal (≥18.5) 47/437 (11) 1
Low (<18.5) 58/310 (19) 1.78 (1.20, 2.64)
HIV status
Negative 6/345 (2) 1
Positive 99/401 (25) 18.18 (7.96, 41.52)
Gender
Female 38/352 (11) 1
Male 67/395 (17) 1.70 (1.12, 2.59)
Smoker2
388
No 84/593 (14) 1
Yes 20/151 (13) 0.62 (0.37, 1.04)
Weight loss3
No 11/165 (7) 1
Yes 94/579 (16) 3.70 (1.97, 6.94)
Chest x-ray extent4
Normal/minimal 19/111 (17) 1
Moderate/far advanced 86/628 (14) 0.73 (0.44, 1.21)
1One missed HIV status; 4eight missed extent variable; 2three missed history of ever smoked; 3four missed history of weight loss; low BMI <18.5 kg/m2.
389
Table 13:7 Relative hazards [HR, 95% confidence intervals (CIs)] for death among tuberculosis patients with normal compared with patients having low body mass index (BMI) stratified according to Age group
Stratified models
Deaths/N Young ≤30 yr Old >30 yr Characteristics (%) (n=434) (n=313)
HR (95% CI) HR (95% CI)
BMI (kg/m2)
Normal (≥18.5) 47/437 (11) 1 1
Low (<18.5) 58/310 (19) 2.85 (1.41, 5.77) 1.39 (0.86, 2.24)
HIV status
Negative 6/345 (2) 1 1
Positive 99/401 (25) 23.94 (5.70, 100.60) 11.75 (4.27, 32.36)
Gender
Female 38/352 (11) 1 1
Male 67/395 (17) 1.15 (0.57, 2.31) 1.88 (1.08, 3.25)
Smoker2
390
No 84/593 (14) 1 1
Yes 20/151 (13) 0.49 (0.17, 1.43) 0.61 (0.34, 1.11)
Weight loss3
No 11/165 (7) 1 1
Yes 94/579 (16) 1.96 (0.84, 4.56) 6.24 (2.27, 17.18)
Chest x-ray extent4
Normal/minimal 19/111 (17) 1 1
Moderate/far advanced 86/628 (14) 0.65 (0.26, 1.61) 0.78 (0.42, 1.43)
1One missed HIV status; 4eight missed extent variable; 2three missed history of ever smoked; 3four missed history of weight loss; low BMI <18.5 kg/m2.
391
Table 13:8 Relative hazards [HR, 95% confidence intervals (CIs)] for death among
HIV positive tuberculosis patients compared with HIV negative patients
Overall model Characteristics Deaths/N (%)
HR (95% CI)
HIV status
Negative 6/345 (2) 1
Positive 99/401 (25) 15.66 (6.84, 35.86)
Age group
≤30 years 34/434 (8) 1
>30 years 71/313 (23) 2.23 (1.47, 3.38)
Gender
Female 38/352 (11) 1
Male 67/395 (17) 1.64 (1.08, 2.50)
Smoker2
No 84/593 (14) 1
Yes 20/151 (13) 0.64 (0.39, 1.07)
392
Weight loss3
No 11/165 (7) 1
Yes 94/579 (16) 3.51 (1.87, 6.59)
Chest x-ray extent4
Normal/minimal 19/111 (17) 1
Moderate/far advanced 86/628 (14) 0.81 (0.49, 1.34)
1One missed HIV status; 4eight missed extent variable; 2three missed history of ever smoked; 3four missed history of weight loss; low BMI <18.5 kg/m2.
393
Table 13:9 Relative hazards [HR, 95% confidence intervals (CIs)] for death among
HIV positive tuberculosis patients compared with HIV negative patients stratified according to BMI categories
Stratified models
Deaths/N Normal BMI Low BMI Characteristics (%) (n=437) (n=310)
HR (95% CI) HR (95% CI)
HIV status
Negative 6/345 (2) 1 1
Positive 99/401 (25) 42.86 (5.90, 311.47) 10.65 (4.20, 27.02)
Age group
≤30 years 34/434 (8) 1 1
>30 years 71/313 (23) 3.10 (1.63, 5.91) 1.68 (0.97, 2.92)
Gender
Female 38/352 (11) 1 1
Male 67/395 (17) 1.51 (0.83, 2.75) 1.59 (0.87, 2.92)
Smoker2
394
No 84/593 (14) 1 1
Yes 20/151 (13) 0.57 (0.23, 1.40) 0.60 (0.32, 1.12)
Weight loss3
No 11/165 (7) 1 1
Yes 94/579 (16) 3.75 (1.48, 9.54) 3.37 (1.43, 7.97)
Chest x-ray extent4
Normal/minimal 19/111 (17) 1 1
Moderate/far advanced 86/628 (14) 0.98 (0.47, 2.03) 0.63 (0.32, 1.25)
1One missed HIV status; 4eight missed extent variable; 2three missed history of ever smoked; 3four missed history of weight loss; low BMI <18.5 kg/m2.
395
Table 13:10 Multivariable Relative hazards [HR, 95% confidence intervals (CIs)] for death among tuberculosis patients with normal compared with patients having low fat-free mass index (FFMI)
Deaths/N Relative 95% Confidence
Characteristic (%) hazard Interval
Fat-free mass index (kg/m2)
Normal 14/208 (7) 1 -
Low 16/103 (16) - -
Sex
Female 12/147 (8) 1 -
Male 18/164 (11) 3.34 1.23, 9.08
Age (years)
≤30 10/194 (5) 1 -
>30 20/117 (17) 2.24 0.84, 5.95
HIV-serostatus1
Negative 1/172 (0.6) 1 -
Positive 29/138 (21) - -
Hemoglobin
396
>10 mg/dl 14/224 (6) 1 -
≤10 mg/dl 15/74 (20) 1.34 0.52, 3.49
Reduced FFMI*sex 0.13 0.03, 0.56
Reduced FFMI*age 1.61 0.35, 7.34
Reduced FFMI*HIV - -
Reduced FFMI*hemoglobin 2.00 0.49, 8.12
-2LL reduced model - -2LL full model: 332.269 – 321.399 = 10.870; df = 4, p=0.028. 1One missed HIV status; 215 missed hemoglobin measurement due to lack of blood; FFMI low (<16.7 kg/m2 for men and <14.6 kg/m2 for women, normal (≥ 16.7kg/m2 for men, ≥14.6 kg/m2 for women).
397
Table 13:11 Relative hazards [HR, 95% confidence intervals (CIs)] for death among tuberculosis patients with normal compared with low fat-free mass index (FFMI)
Overall model Characteristics Deaths/N (%)
HR (95% CI)
FFMI in kg/m2
Normal (≥18.5) 14/208 (7) 1
Low (<18.5) 16/103 (16) 1.50 (0.73, 3.08)
Gender
Female 12/147 (8) 1
Male 18/164 (11) 1.55 (0.72, 3.37)
Age (years)
≤30 10/194 (5) 1
>30 20/117 (17) 3.57 (1.73, 7.35)
Hemoglobin
2 (mg/dl)
>10 14/224 (6) 1
398
≤10 15/74 (20) 2.85 (1.45, 5.61)
1One missed HIV status; 215 missed hemoglobin measurement due to lack of blood; FFMI low (<16.7 kg/m2 for men and <14.6 kg/m2 for women, normal (≥ 16.7kg/m2 for men, ≥14.6 kg/m2 for women).
399
Table 13:12 Relative hazards [HR, 95% confidence intervals (CIs)] for death among tuberculosis patients with normal compared with low fat-free mass index (FFMI) stratified according to HIV status
Stratified models
HIV negative HIV positive Characteristics Deaths/N (%) (n=186) (n=153)
HR (95% CI) HR (95% CI)
FFMI in kg/m2
Normal (≥18.5) 14/208 (7) 1 1
Low (<18.5) 16/103 (16) - 1.50 (0.72, 3.12)
Gender
Female 12/147 (8) 1 1
Male 18/164 (11) - 1.53 (0.69, 3.39)
Age (years)
≤30 10/194 (5) 1 1
>30 20/117 (17) - 2.21 (1.06, 4.60)
Hemoglobin
2 (mg/dl)
400
>10 14/224 (6) 1 1
≤10 15/74 (20) - 1.65 (0.82, 3.30)
1One missed HIV status; 215 missed hemoglobin measurement due to lack of blood; FFMI low (<16.7 kg/m2 for men and <14.6 kg/m2 for women, normal (≥ 16.7kg/m2 for men, ≥14.6 kg/m2 for women).
401
Table 13:13 Relative hazards [HR, 95% confidence intervals (CIs)] for death among tuberculosis patients with normal compared with low fat-free mass index (FFMI)
Overall model Characteristics Deaths/N (%)
HR (95% CI)
FFMI in kg/m2
Normal (≥18.5) 14/208 (7) 1
Low (<18.5) 16/103 (16) 1.55 (0.76, 3.15)
Gender
Female 12/147 (8) 1
Male 18/164 (11) 1.92 (0.90, 4.11)
HIV-serostatus1
Negative 1/172 (0.6) 1
Positive 29/138 (21) 41.68 (5.67, 306.41)
Hemoglobin
2 (mg/dl)
>10 14/224 (6) 1
402
≤10 15/74 (20) 1.79 (0.90, 3.55)
1One missed HIV status; 215 missed hemoglobin measurement due to lack of blood; FFMI low (<16.7 kg/m2 for men and <14.6 kg/m2 for women, normal (≥ 16.7kg/m2 for men, ≥14.6 kg/m2 for women).
403
Table 13:14 Relative hazards [HR, 95% confidence intervals (CIs)] for death among
tuberculosis patients with normal compared with low fat-free mass index (FFMI) stratified according to age group
Stratified models
Young ≤30 years Old >30 years Characteristics Deaths/N (%) (n=212) (n=128)
HR (95% CI) HR (95% CI)
FFMI in kg/m2
Normal (≥18.5) 14/208 (7) 1 1
Low (<18.5) 16/103 (16) 1.89 (0.55, 6.52) 1.52 (0.63, 3.69)
Gender
Female 12/147 (8) 1 1
Male 18/164 (11) 1.62 (0.45, 5.80) 1.56 (0.58, 4.19)
HIV-serostatus1
Negative 1/172 (0.6) 1 1
Positive 29/138 (21) - 16.03 (2.14, 119.84)
Hemoglobin
2 (mg/dl)
404
>10 14/224 (6) 1 1
≤10 15/74 (20) 1.43 (0.42, 4.91) 2.02 (0.89, 4.58)
1One missed HIV status; 215 missed hemoglobin measurement due to lack of blood; FFMI low (<16.7 kg/m2 for men and <14.6 kg/m2 for women, normal (≥ 16.7kg/m2 for men, ≥14.6 kg/m2 for women).
405
Table 13:15 Relative hazards [HR, 95% confidence intervals (CIs)] for death among
HIV positive tuberculosis patients compared with HIV negative patients
Overall model Characteristics Deaths/N (%)
HR (95% CI)
HIV status
Negative 1/172 (0.6) 1
Positive 29/138 (21) 34.37 (4.65, 254.20)
Age (years)
≤30 10/194 (5) 1
>30 20/117 (17) 2.33 (1.12, 4.83)
Gender
Female 12/147 (8) 1
Male 18/164 (11) 1.89 (0.91, 3.91)
Hemoglobin
2 (mg/dl)
>10 14/224 (6) 1
406
≤10 15/74 (20) 1.86 (0.94, 3.67)
1One missed HIV status; 215 missed hemoglobin measurement due to lack of blood; FFMI low (<16.7 kg/m2 for men and <14.6 kg/m2 for women, normal (≥ 16.7kg/m2 for men, ≥14.6 kg/m2 for women).
407
Table 13:16 Relative hazards [HR, 95% confidence intervals (CIs)] for death among
HIV positive tuberculosis patients compared with HIV negative patients stratified according to fat-free mass index (FFMI) categories
Stratified models
Normal FFMI Characteristics Deaths/N (%) Low FFMI (n=103) (n=208)
HR (95% CI) HR (95% CI)
HIV status
Negative 1/172 (0.6) 1 1
Positive 29/138 (21) - 14.80 (1.93, 113.58)
Age (years)
≤30 10/194 (5) 1 1
>30 20/117 (17) 2.04 (0.68, 6.10) 3.50 (1.10, 11.11)
Gender
Female 12/147 (8) 1 1
Male 18/164 (11) 2.28 (0.77, 6.75) 0.43 (0.14, 1.31)
Hemoglobin
2 (mg/dl)
408
>10 14/224 (6) 1 1
≤10 15/74 (20) 1.50 (0.51, 4.42) 2.64 (0.95, 7.39)
1One missed HIV status; 215 missed hemoglobin measurement due to lack of blood; FFMI low (<16.7 kg/m2 for men and <14.6 kg/m2 for women, normal (≥ 16.7kg/m2 for men, ≥14.6 kg/m2 for women).
409
Figure 13:1 Survival distribution among women with low baseline body mass index
(BMI) (<18.5 kg/m2) compared to women with normal BMI in Uganda
410
Figure 13:2 Survival distribution among men with low (<16.7 kg/m2 for men, <14.6 kg/m2 for women) baseline fat-free mass index (FFMI) compared to men with normal FFMI in Uganda
411
Figure 13:3 Survival distribution among Men compared to Women tuberculosis patients in urban Uganda
Data truncated at 1280 days of follow-up
412
Figure 13:4 Survival distribution among Men compared with Women among HIV negative tuberculosis patients in Uganda
413
CHAPTER 14
EXTRA TABLES AND FIGURES FOR CHAPTER TEN
Table 14:1 Spearman raw correlation matrix for fat-free mass, fat mass, and body mass index over different visit time points
414
Visit
(Month)
Fat-free mass
0 Fat-free mass 1
3 Fat-free mass 0.73 1
12 Fat-free mass 0.69 0.78 1
24 Fat-free mass 0.68 0.72 0.78
Fat mass
0 Fat mass 1
3 Fat mass 0.88 1
12 Fat mass 0.85 0.87 1
24 Fat mass 0.81 0.85 0.90
Body mass index
0 Body mass index 1
2 Body mass index 0.83 1
3 Body mass index 0.88 0.91 1
5 Body mass index 0.82 0.87 0.90 1
415
6 Body mass index 0.68 0.77 077 0.77 1
12 Body mass index 0.82 0.83 0.86 0.85 0.74 1
24 Body mass index 0.75 0.80 0.80 0.83 0.71 0.89
Table 14:2 Regression estimates for effects of baseline wasting, lag FFMI measure, and baseline characteristics on probability of missing
416
Covariate Estimate Standard error 95% CI
Proc GENMOD FFMI
Intercept -1.04 1.79 - 4.56, 2.48
Baseline wasting -0.34 0.41 -1.25, 0.47
Baseline FFMI -0.03 0.12 -0.26, 0.20
Lag FFMI -0.11 0.10 -1.07, 0.29
Sex 0.28 0.28 -0.27, 0.82
Age 0.15 0.20 -0.25, 0.54
HIV 0.42 0.20 2.10, 0.04
Smoker 0.29 0.26 -0.23, 0.81
Hemoglobin -0.42 0.23 -0.87, 0.02
History of weight loss 0.09 0.25 -0.40, 0.58
Chest x-ray extent -0.11 0.27 -0.63, 0.41
Time*baseline wasting 0.02 0.02 -0.02, 0.06
Proc MIXED FFMI
Intercept 15.65 0.50 14.67, 16.63
Probability missing -0.06 0.12 -0.30, 0.18
417
Baseline wasting -1.68 0.14 -1.95, -1.41
Lag FFMI 0.04 0.03 -0.02, 0.09
Sex 1.74 0.13 1.48, 2.00
Age 0.10 0.12 -0.13, 0.34
HIV 0.06 0.12 -0.18, 0.29
Smoker -0.14 0.15 -0.44, 0.17
Hemoglobin -0.34 0.14 -0.60, -0.07
History of weight loss -0.03 0.15 -0.33, 0.28
Chest x-ray extent -0.10 0.16 -0.40, 0.21
Time*missing -0.01 0.01 -0.02, 0.01
Random intercept (b0i) variance 0.458 0.098 <0.001*
Random slope (b1i) variance 0.007 0.004 0.133*
Covariance slopes (b0i, b1i) 0.0004 0.0003 0.097*
Autocorrelation; exponential 0.8610 0 -
Residual (measurement error) 0.567 0.053 <0.001*
Table 14:3 Regression estimates for effects of baseline wasting, lag FMI measure, and baseline characteristics on probability of missing
418
Covariate Estimate Standard error 95% CI
Proc GENMOD FMI
Intercept -3.84 0.63 - 5.08, -2.59
Baseline wasting 0.45 0.40 -0.34, 1.24
Baseline FMI -0.10 0.07 -0.03, 0.24
Lag FMI -0.04 0.05 -0.14, 0.06
Sex 0.25 0.31 -0.35, 0.84
Age 0.16 0.20 -0.23, 0.55
HIV 0.44 0.20 0.05, 0.83
Smoker 0.32 0.27 -0.20, 0.84
Hemoglobin -0.29 0.24 -0.76, 0.17
History of weight loss 0.17 0.26 -0.34, 0.67
Chest x-ray extent -0.01 0.27 -0.53, 0.51
Time*baseline wasting -0.02 0.02 -0.06, 0.03
Proc MIXED FMI
Intercept 6.37 0.36 5.66, 7.08
Probability missing 0.31 0.20 -0.08, 0.70
419
Baseline wasting -2.18 0.20 -2.57, -1.79
Lag FMI 0.03 0.02 -0.01, 0.07
Sex -2.91 0.21 -3.32, -2.49
Age -0.02 0.20 -0.42, 0.38
HIV -0.14 0.20 -0.54, 0.25
Smoker 0.24 0.26 -0.26, 0.75
Hemoglobin -0.44 0.24 -0.90, 0.03
History of weight loss -0.52 0.26 -1.03, -0.02
Chest x-ray extent -0.45 0.26 -0.96, 0.06
Time*missing 0.005 0.01 -0.02, 0.03
Random intercept (b0i) 1.721 0.207 <0.001* variance
Random slope (b1i) 0.004 0.009 0.642* variance
Covariance slopes (b0i, b1i) 0.003 0.0006 <0.001*
Autocorrelation; 0.9997 3.383E8 0.500* exponential
0.684 0.061 <0.001* Residual (measurement
420
error)
*p-values
421
Table 14:4 Assessing contributions of polynomials, random intercepts, and random slopes in mixed models
Fat-free mass
Parms Dif - p-
Model AIC -2RLL (df+1) 2RLL df value
Linear model 2768.2 2762.2 3 - - -
Quadratic model 2751.4 2745.4 3 16.8
Cubic model 2730.3 2724.3 3 37.9
Fixed intercepts 2691.6 2673.6 3 - - -
Random 0 0 NA intercepts 2691.6 2673.6 3
Random
Intercepts and 27 2 <0.001 slopes 2668.6 2646.6 5
Fat mass
Linear model 3197.6 3191.6 3 - - -
Quadratic model 3182.0 3176.0 3 15.6 0 NA
Cubic model 3190.5 3184.5 3 7.1 0 NA
422
Fixed intercepts 3155.6 3137.6 3 - - -
Random 0 intercepts 3155.6 3137.6 3
Random
intercepts and 52.2 2 <0.001 slopes 3107.4 3085.4 5
Body mass index
Linear model 15923.8 15917.8 3 - - -
Quadratic model 15768.2 15762.2 3 155.6 0 NA
Cubic model 15688.8 15682.8 3 235.0 0 NA
Fixed intercepts 15650.0 15632.0 3 - - -
Random 5.6 1 intercepts 15646.4 15626.4 4
Random intercepts and 116.5 3 <0.001 slopes 15539.5 15515.5 6
AIC = Akaike Information Criteria, -2LL = -2 log likelihood, df = degrees of freedom, NA = not applicable.
423
Table 14:5 Assessing Piecewise and polynomial models; random intercepts, and random slopes in mixed models
Fat-free mass
Parms Dif - p-
Model AIC -2RLL (df+1) 2RLL df value
Linear model 2779.4 2767.4 6 - - -
Quadratic model 2755.6 2737.6 9 29.8 3 <0.001
2 Spline model 2714.8 2686.8 15 NA NA NA
Random intercepts 2734.7 2722.7 6 - - -
Random early slope 2715.8 2699.8 8 22.9 2 <0.001
Random early & 3 months slopes 2709.5 2687.5 11 35.2 5 <0.001
Random early, 3, &
12 months slopes 2714.8 2686.8 15 35.9 9 <0.001
Fat mass
Linear model 3210.4 3202.4 6 - - -
Quadratic model 3276.4 3262.4 9 60 3 <0.001
424
2 Spline model 3274.8 3252.8 15 NA NA NA
Random intercepts 3357.2 3361.2 6 - - -
Random early slope 3273.2 3265.2 8 96 2 <0.001
Random early & 3 months slopes 3276.4 3264.4 11 96.8 4 <0.001
Random early, 3, &
12 months slopes 3274.8 3252.8 15 108.4 9 <0.001
Body mass index
Linear model 16307.3 16295.3 6 - - -
Quadratic model 16003.5 15983.5 9 311.8 3 <0.001
2 Spline model 15887.2 15859.2 15 NA NA NA
Random intercepts 16079.2 16067.2 6 - - -
Random early slope 15911.8 15895.8 8 171.4 2 <0.001
Random early & 3 months slopes 15894.2 15872.2 11 195.0 5 <0.001
Random early, 3, &
12 months slopes 15887.2 15859.2 15 208.0 8 <0.001
AIC = Akaike Information Criteria, -2LL = -2 log likelihood, df = degrees of freedom, NA = not applicable.
425
Figure 14:1 Individual profiles for fat-free mass over time
426
Figure 14:2 Individual profiles for fat mass index over time
427
Figure 14:3 Individual profiles for BMI over time
428
Figure 14:4 Box plots for fat-free mass index over time
429
Figure 14:5 Box plots for fat mass index over time
430
Figure 14:6 Box plots for BMI over time
431
Figure 14:7 Mean profile for FFMI among patients with compared to patients without baseline wasting
432
Figure 14:8 Mean profile for FFMI among men compared to mean profile among women
433
Figure 14:9 Mean profile for FFMI among HIV negative compared to mean profile among HIV positive patients
434
Figure 14:10 Mean profile for FMI among patients with compared to patients without wasting
435
Figure 14:11 Mean profile for FMI among men compared to mean profile among women
436
Figure 14:12 Mean profile for FMI among HIV negative compared to mean profile among HIV positive patients
437
Figure 14:13 Mean profile for BMI among patients with compared to patients without wasting
438
Figure 14:14 Mean profiles for BMI among men compared to mean profile among women
439
Figure 14:15 Mean profile for BMI among HIV negative compared to mean profile among HIV positive patients
440
BIBLIOGRAPHY
CHAPTER ONE
Chandra, R. K. 1991. 1990 McCollum Award lecture. Nutrition and immunity: lessons
from the past and new insights into the future. Am J Clin Nutr 53 (5):1087-101.
Fernandes G, Jolly C.A, and Lawrence R.A. 2006. Nutrition and The Immune System. In
Modern Nutrition in Health and Disease, edited by Shills M.E, Shike M, Ross
C.A, Caballero B and Cousins R.J. Philadelphia PA: Lippincott Williams &
Wilkins.
Gershwin M.E, Beach R.S, and Hurley L.S. 1985. Nutrition and Immunity. Olando FL:
Academic Press.
Harries, A. D., W. A. Nkhoma, P. J. Thompson, D. S. Nyangulu, and J. J. Wirima. 1988.
Nutritional status in Malawian patients with pulmonary tuberculosis and response
to chemotherapy. Eur J Clin Nutr 42 (5):445-50.
Kennedy, N., A. Ramsay, L. Uiso, J. Gutmann, F. I. Ngowi, and S. H. Gillespie. 1996.
Nutritional status and weight gain in patients with pulmonary tuberculosis in
Tanzania. Trans R Soc Trop Med Hyg 90 (2):162-6.
Kyle, U. G., I. Bosaeus, A. D. De Lorenzo, P. Deurenberg, M. Elia, J. Manuel Gomez, B.
Lilienthal Heitmann, L. Kent-Smith, J. C. Melchior, M. Pirlich, H. Scharfetter, M.
W. J. Schols A, and C. Pichard. 2004. Bioelectrical impedance analysis-part II:
utilization in clinical practice. Clin Nutr 23 (6):1430-53.
441
Kyle, U. G., L. Genton, and C. Pichard. 2002. Body composition: what's new? Curr Opin
Clin Nutr Metab Care 5 (4):427-33.
Kyle, U. G., A. Piccoli, and C. Pichard. 2003. Body composition measurements:
interpretation finally made easy for clinical use. Curr Opin Clin Nutr Metab Care
6 (4):387-93.
Lawn, S. D., and G. Churchyard. 2009. Epidemiology of HIV-associated tuberculosis.
Curr Opin HIV AIDS 4 (4):325-33.
Lucas, S. B., K. M. De Cock, A. Hounnou, C. Peacock, M. Diomande, M. Honde, A.
Beaumel, L. Kestens, and A. Kadio. 1994. Contribution of tuberculosis to slim
disease in Africa. Bmj 308 (6943):1531-3.
Lucas, S. B., A. Hounnou, C. Peacock, A. Beaumel, G. Djomand, J. M. N'Gbichi, K.
Yeboue, M. Honde, M. Diomande, C. Giordano, and et al. 1993. The mortality
and pathology of HIV infection in a west African city. Aids 7 (12):1569-79.
Mostert, R., A. Goris, C. Weling-Scheepers, E. F. Wouters, and A. M. Schols. 2000.
Tissue depletion and health related quality of life in patients with chronic
obstructive pulmonary disease. Respir Med 94 (9):859-67.
Mugusi, F. M., S. Mehta, E. Villamor, W. Urassa, E. Saathoff, R. J. Bosch, and W. W.
Fawzi. 2009. Factors associated with mortality in HIV-infected and uninfected
patients with pulmonary tuberculosis. BMC Public Health 9:409.
Mupere, E., S. Zalwango, A. Chiunda, A. Okwera, R. Mugerwa, and C. Whalen. 2010.
Body composition among HIV-seropositive and HIV-seronegative adult patients
with pulmonary tuberculosis in Uganda. Ann Epidemiol 20 (3):210-6.
442
Niyongabo, T., D. Henzel, M. Idi, S. Nimubona, E. Gikoro, J. C. Melchior, S. Matheron,
G. Kamanfu, B. Samb, B. Messing, J. Begue, P. Aubry, and B. Larouze. 1999.
Tuberculosis, human immunodeficiency virus infection, and malnutrition in
Burundi. Nutrition 15 (4):289-93.
Nunn, P., R. Brindle, L. Carpenter, J. Odhiambo, K. Wasunna, R. Newnham, W. Githui,
S. Gathua, M. Omwega, and K. McAdam. 1992. Cohort study of human
immunodeficiency virus infection in patients with tuberculosis in Nairobi, Kenya.
Analysis of early (6-month) mortality. Am Rev Respir Dis 146 (4):849-54.
Paton, N. I., and Y. M. Ng. 2006. Body composition studies in patients with wasting
associated with tuberculosis. Nutrition 22 (3):245-51.
Ramakrishnan, C. V., K. Rajendran, P. G. Jacob, W. Fox, and S. Radhakrishna. 1961.
The role of diet in the treatment of pulmonary tuberculosis. An evaluation in a
controlled chemotherapy study in home and sanatorium patients in South India.
Bull World Health Organ 25:339-59.
Scalcini, M., R. Occenac, J. Manfreda, and R. Long. 1991. Pulmonary tuberculosis,
human immunodeficiency virus type-1 and malnutrition. Bull Int Union Tuberc
Lung Dis 66 (1):37-41.
Suttmann, U., J. Ockenga, O. Selberg, L. Hoogestraat, H. Deicher, and M. J. Muller.
1995. Incidence and prognostic value of malnutrition and wasting in human
immunodeficiency virus-infected outpatients. J Acquir Immune Defic Syndr Hum
Retrovirol 8 (3):239-46.
443
van Lettow, M., W. W. Fawzi, and R. D. Semba. 2003. Triple trouble: the role of
malnutrition in tuberculosis and human immunodeficiency virus co-infection.
Nutr Rev 61 (3):81-90.
Vitoria, M., R. Granich, C. F. Gilks, C. Gunneberg, M. Hosseini, W. Were, M.
Raviglione, and K. M. De Cock. 2009. The global fight against HIV/AIDS,
tuberculosis, and malaria: current status and future perspectives. Am J Clin Pathol
131 (6):844-8.
Wagner, G. J., S. J. Ferrando, and J. G. Rabkin. 2000. Psychological and physical health
correlates of body cell mass depletion among HIV+ men. J Psychosom Res 49
(1):55-7.
World Health Organization. Global tuberculosis control. Epidemiology, strategy,
financing. WHO/HTM/TB/2009.411. 2009. Geneva: World Health Organization
Zachariah, R., M. P. Spielmann, A. D. Harries, and F. M. Salaniponi. 2002. Moderate to
severe malnutrition in patients with tuberculosis is a risk factor associated with
early death. Trans R Soc Trop Med Hyg 96 (3):291-4.
444
CHAPTER TWO
Arkhipova O.P. 1975. Impact of tuberculosis infection and antibacterial preparations on
thiamin metabolism. Voprosy Meditsinskoi Khimii 21:449 - 560.
Baum, M. K., E. Mantero-Atienza, G. Shor-Posner, M. A. Fletcher, R. Morgan, C.
Eisdorfer, H. E. Sauberlich, P. E. Cornwell, and R. S. Beach. 1991. Association of
vitamin B6 status with parameters of immune function in early HIV-1 infection. J
Acquir Immune Defic Syndr 4 (11):1122-32.
Beach, R. S., E. Mantero-Atienza, G. Shor-Posner, J. J. Javier, J. Szapocznik, R. Morgan,
H. E. Sauberlich, P. E. Cornwell, C. Eisdorfer, and M. K. Baum. 1992. Specific
nutrient abnormalities in asymptomatic HIV-1 infection. Aids 6 (7):701-8.
Bhaskaram, P. 1992. The vicious cycle of malnutrition-infection with special reference to
diarrhea, measles and tuberculosis. Indian Pediatr 29 (6):805-14.
Bhuyan, U. N., and V. Ramalingaswami. 1973. Immune responses of the protein-
deficient guinea pig to BCG vaccination. Am J Pathol 72 (3):489-502.
Bogden, J. D., H. Baker, O. Frank, G. Perez, F. Kemp, K. Bruening, and D. Louria. 1990.
Micronutrient status and human immunodeficiency virus (HIV) infection. Ann N
Y Acad Sci 587:189-95.
Cegielski, J. P., and D. N. McMurray. 2004. The relationship between malnutrition and
tuberculosis: evidence from studies in humans and experimental animals. Int J
Tuberc Lung Dis 8 (3):286-98.
445
Chan J, Tanaka K.E, Mannion C, Carroll D, Tsang M.S, Xing Y, Lowenstein C, and
Bloom B.R. 1997. Effects of protein calorie malnutrition on mice infected with
BCG. J Nutr Immunol.
Chandra, R. K. 1991. 1990 McCollum Award lecture. Nutrition and immunity: lessons
from the past and new insights into the future. Am J Clin Nutr 53 (5):1087-101.
Chandra, R. K., and S. Kumari. 1994. Nutrition and immunity: an overview. J Nutr 124
(8 Suppl):1433S-1435S.
Comstock, G. W., and C. E. Palmer. 1966. Long-term results of BCG vaccination in the
southern United States. Am Rev Respir Dis 93 (2):171-83.
Ehrenpreis, E. D., S. J. Carlson, H. L. Boorstein, and R. M. Craig. 1994. Malabsorption
and deficiency of vitamin B12 in HIV-infected patients with chronic diarrhea. Dig
Dis Sci 39 (10):2159-62.
Evans, D. I., and B. Attock. 1971. Folate deficiency in pulmonary tuberculosis:
relationship to treatment and to serum vitamin A and beta-carotene. Tubercle 52
(4):288-94.
Harries, A. D., W. A. Nkhoma, P. J. Thompson, D. S. Nyangulu, and J. J. Wirima. 1988.
Nutritional status in Malawian patients with pulmonary tuberculosis and response
to chemotherapy. Eur J Clin Nutr 42 (5):445-50.
Harriman, G. R., P. D. Smith, M. K. Horne, C. H. Fox, S. Koenig, E. E. Lack, H. C.
Lane, and A. S. Fauci. 1989. Vitamin B12 malabsorption in patients with acquired
immunodeficiency syndrome. Arch Intern Med 149 (9):2039-41.
446
Karyadi, E., W. Schultink, R. H. Nelwan, R. Gross, Z. Amin, W. M. Dolmans, J. W. van
der Meer, J. G. Hautvast, and C. E. West. 2000. Poor micronutrient status of
active pulmonary tuberculosis patients in Indonesia. J Nutr 130 (12):2953-8.
Karyadi, E., C. E. West, W. Schultink, R. H. Nelwan, R. Gross, Z. Amin, W. M.
Dolmans, H. Schlebusch, and J. W. van der Meer. 2002. A double-blind, placebo-
controlled study of vitamin A and zinc supplementation in persons with
tuberculosis in Indonesia: effects on clinical response and nutritional status. Am J
Clin Nutr 75 (4):720-7.
Keusch, G. T. 1990. Micronutrients and susceptibility to infection. Ann N Y Acad Sci
587:181-8.
Line, D. H., B. Seitanidis, J. O. Morgan, and A. V. Hoffbrand. 1971. The effects of
chemotherapy on iron, folate, and vitamin B 12 metabolism in tuberculosis. Q J
Med 40 (159):331-40.
Lucas, S. B., K. M. De Cock, A. Hounnou, C. Peacock, M. Diomande, M. Honde, A.
Beaumel, L. Kestens, and A. Kadio. 1994. Contribution of tuberculosis to slim
disease in Africa. Bmj 308 (6943):1531-3.
Macallan, D. C. 1999. Malnutrition in tuberculosis. Diagn Microbiol Infect Dis 34
(2):153-7.
Macallan, D. C., M. A. McNurlan, A. V. Kurpad, G. de Souza, P. S. Shetty, A. G. Calder,
and G. E. Griffin. 1998. Whole body protein metabolism in human pulmonary
tuberculosis and undernutrition: evidence for anabolic block in tuberculosis. Clin
Sci (Lond) 94 (3):321-31.
447
Macallan, D. C., M. A. McNurlan, E. Milne, A. G. Calder, P. J. Garlick, and G. E.
Griffin. 1995. Whole-body protein turnover from leucine kinetics and the
response to nutrition in human immunodeficiency virus infection. Am J Clin Nutr
61 (4):818-26.
Macallan, D. C., C. Noble, C. Baldwin, S. A. Jebb, A. M. Prentice, W. A. Coward, M. B.
Sawyer, T. J. McManus, and G. E. Griffin. 1995. Energy expenditure and wasting
in human immunodeficiency virus infection. N Engl J Med 333 (2):83-8.
Madebo, T., G. Nysaeter, and B. Lindtjorn. 1997. HIV infection and malnutrition change
the clinical and radiological features of pulmonary tuberculosis. Scand J Infect
Dis 29 (4):355-9.
Markkanen, T., A. Levanto, V. Sallinen, and S. Virtanen. 1967. Folic acid and vitamin
B12 in tuberculosis. Scand J Haematol 4 (4):283-91.
McMurray, D. N., R. A. Bartow, C. L. Mintzer, and E. Hernandez-Frontera. 1990.
Micronutrient status and immune function in tuberculosis. Ann N Y Acad Sci
587:59-69.
Miansikov V.G. 1969. Some indices of vitamion B6 metabolism in patients with
pulmonary tuberculosis in elderly and old persons. Vrachebnoe Delo 3:77 - 79.
Miller, L. G., S. M. Asch, E. I. Yu, L. Knowles, L. Gelberg, and P. Davidson. 2000. A
population-based survey of tuberculosis symptoms: how atypical are atypical
presentations? Clin Infect Dis 30 (2):293-9.
Mupere, E., S. Zalwango, A. Chiunda, A. Okwera, R. Mugerwa, and C. Whalen. 2010.
Body composition among HIV-seropositive and HIV-seronegative adult patients
with pulmonary tuberculosis in Uganda. Ann Epidemiol 20 (3):210-6.
448
Onwubalili, J. K. 1988. Malnutrition among tuberculosis patients in Harrow, England.
Eur J Clin Nutr 42 (4):363-6.
Panasyuk A.V, Penenko O.R, Kuzmenko I.V, and et al. 1991. Vitamin E and its structural
analogs in tuberculosis. Urkr Biokhim Zh 63:83 - 88.
Paton, N. I., L. R. Castello-Branco, G. Jennings, M. B. Ortigao-de-Sampaio, M. Elia, S.
Costa, and G. E. Griffin. 1999. Impact of tuberculosis on the body composition of
HIV-infected men in Brazil. J Acquir Immune Defic Syndr Hum Retrovirol 20
(3):265-71.
Paton, N. I., Y. K. Chua, A. Earnest, and C. B. Chee. 2004. Randomized controlled trial
of nutritional supplementation in patients with newly diagnosed tuberculosis and
wasting. Am J Clin Nutr 80 (2):460-5.
Paton, N. I., and Y. M. Ng. 2006. Body composition studies in patients with wasting
associated with tuberculosis. Nutrition 22 (3):245-51.
Paton, N. I., Y. M. Ng, C. B. Chee, C. Persaud, and A. A. Jackson. 2003. Effects of
tuberculosis and HIV infection on whole-body protein metabolism during feeding,
measured by the [15N]glycine method. Am J Clin Nutr 78 (2):319-25.
Rao K.N, and Gopalan C. 1966. The role of nutritional factors in tuberculosis. Indian J
Tuberculosis 13:102 - 106.
Rook, G. A., and R. Hernandez-Pando. 1996. The pathogenesis of tuberculosis. Annu Rev
Microbiol 50:259-84.
Safarian, M. D., K. G. Karagezian, E. T. Karapetian, and N. A. Avanesian. 1990. [The
efficacy of antioxidant therapy in patients with tuberculosis of the lungs and the
correction of lipid peroxidation processes]. Probl Tuberk (5):40-4.
449
Sarraf, P., R. C. Frederich, E. M. Turner, G. Ma, N. T. Jaskowiak, D. J. Rivet, 3rd, J. S.
Flier, B. B. Lowell, D. L. Fraker, and H. R. Alexander. 1997. Multiple cytokines
and acute inflammation raise mouse leptin levels: potential role in inflammatory
anorexia. J Exp Med 185 (1):171-5.
Scalcini, M., R. Occenac, J. Manfreda, and R. Long. 1991. Pulmonary tuberculosis,
human immunodeficiency virus type-1 and malnutrition. Bull Int Union Tuberc
Lung Dis 66 (1):37-41.
Schwenk A, and Macallan D.C. 2000. Tuberculosis, malnutrition and wasting. Curr Opin
Clin Nutr Metab Care 3:285 - 91.
Schwenk, A., L. Hodgson, A. Wright, L. C. Ward, C. F. Rayner, S. Grubnic, G. E.
Griffin, and D. C. Macallan. 2004. Nutrient partitioning during treatment of
tuberculosis: gain in body fat mass but not in protein mass. Am J Clin Nutr 79
(6):1006-12.
Semba, R. D., and A. M. Tang. 1999. Micronutrients and the pathogenesis of human
immunodeficiency virus infection. Br J Nutr 81 (3):181-9.
Smurova, T. F., and D. I. Prokop'ev. 1969. [Vitamin A and carotene levels in the blood of
pati-ents with pulmonary tuberculosis and diabetes mellitus]. Probl Tuberk 47
(11):50-5.
Tverdal, A. 1986. Body mass index and incidence of tuberculosis. Eur J Respir Dis 69
(5):355-62.
Ullrich, R., T. Schneider, W. Heise, W. Schmidt, R. Averdunk, E. O. Riecken, and M.
Zeitz. 1994. Serum carotene deficiency in HIV-infected patients. Berlin
Diarrhoea/Wasting Syndrome Study Group. Aids 8 (5):661-5.
450
van Lettow, M., W. W. Fawzi, and R. D. Semba. 2003. Triple trouble: the role of
malnutrition in tuberculosis and human immunodeficiency virus co-infection.
Nutr Rev 61 (3):81-90.
Verbon, A., N. Juffermans, S. J. Van Deventer, P. Speelman, H. Van Deutekom, and T.
Van Der Poll. 1999. Serum concentrations of cytokines in patients with active
tuberculosis (TB) and after treatment. Clin Exp Immunol 115 (1):110-3.
Volosevich, G. V. 1982. [Blood T-lymphocyte functional activity in pulmonary
tuberculosis patients undergoing chemotherapeutic agent and vitamin treatment].
Probl Tuberk (3):47-50.
Whalen C, and Semba R.D. 2001. Tuberculosis. In Nutrition and Health in Developing
Countries, edited by Semba R.D and Bloem M.W. Totowa NJ, Human Press.
Zacharia R, M. P Spielmann, Harries A.D, and Salaniponi F.M. 2002. Malnutrition in
tuberculosis patients on admission and weight-gain in relation to HIV status in
Thyolo district of Malawi. Malawi Med J 13:12 - 13.
Zachariah, R., M. P. Spielmann, A. D. Harries, and F. M. Salaniponi. 2002. Moderate to
severe malnutrition in patients with tuberculosis is a risk factor associated with
early death. Trans R Soc Trop Med Hyg 96 (3):291-4.
451
CHAPTER THREE
Guwatudde, D., M. Nakakeeto, E. C. Jones-Lopez, A. Maganda, A. Chiunda, R. D.
Mugerwa, J. J. Ellner, G. Bukenya, and C. C. Whalen. 2003. Tuberculosis in
household contacts of infectious cases in Kampala, Uganda. Am J Epidemiol 158
(9):887-98.
International Union Against Tuberculosis and Lung Disease. Technical guide for sputum
examination for tuberculosis by direct microscopy. 1986. Bull Int Union Tuberc
Lung Dis 61:1 - 16.
Jackson, A. S., M. L. Pollock, J. E. Graves, and M. T. Mahar. 1988. Reliability and
validity of bioelectrical impedance in determining body composition. J Appl
Physiol 64 (2):529-34.
Kotler, D. P., S. Burastero, J. Wang, and R. N. Pierson, Jr. 1996. Prediction of body cell
mass, fat-free mass, and total body water with bioelectrical impedance analysis:
effects of race, sex, and disease. Am J Clin Nutr 64 (3 Suppl):489S-497S.
Kyle, U. G., I. Bosaeus, A. D. De Lorenzo, P. Deurenberg, M. Elia, J. Manuel Gomez, B.
Lilienthal Heitmann, L. Kent-Smith, J. C. Melchior, M. Pirlich, H. Scharfetter, M.
W. J. Schols A, and C. Pichard. 2004. Bioelectrical impedance analysis-part II:
utilization in clinical practice. Clin Nutr 23 (6):1430-53.
Kyle, U. G., L. Genton, and C. Pichard. 2002. Body composition: what's new? Curr Opin
Clin Nutr Metab Care 5 (4):427-33.
452
Kyle, U. G., A. Piccoli, and C. Pichard. 2003. Body composition measurements:
interpretation finally made easy for clinical use. Curr Opin Clin Nutr Metab Care
6 (4):387-93.
Mayanja-Kizza, H., E. Jones-Lopez, A. Okwera, R. S. Wallis, J. J. Ellner, R. D.
Mugerwa, and C. C. Whalen. 2005. Immunoadjuvant prednisolone therapy for
HIV-associated tuberculosis: a phase 2 clinical trial in Uganda. J Infect Dis 191
(6):856-65.
Physical status: the use and interpretation of anthropometry. Report of a WHO Expert
Committee. World Health Organ Tech Rep Ser, 854: 1 - 452. 1995. Geneva.
Schutz, Y., U. U. Kyle, and C. Pichard. 2002. Fat-free mass index and fat mass index
percentiles in Caucasians aged 18-98 y. Int J Obes Relat Metab Disord 26
(7):953-60.
Shah, S., C. Whalen, D. P. Kotler, H. Mayanja, A. Namale, G. Melikian, R. Mugerwa,
and R. D. Semba. 2001. Severity of human immunodeficiency virus infection is
associated with decreased phase angle, fat mass and body cell mass in adults with
pulmonary tuberculosis infection in Uganda. J Nutr 131 (11):2843-7.
Stein, C. M., L. Nshuti, A. B. Chiunda, W. H. Boom, R. C. Elston, R. D. Mugerwa, S. K.
Iyengar, and C. C. Whalen. 2005. Evidence for a major gene influence on tumor
necrosis factor-alpha expression in tuberculosis: path and segregation analysis.
Hum Hered 60 (2):109-18.
Van Lettow, M., J. J. Kumwenda, A. D. Harries, C. C. Whalen, T. E. Taha, N.
Kumwenda, C. Kang'ombe, and R. D. Semba. 2004. Malnutrition and the severity
453
of lung disease in adults with pulmonary tuberculosis in Malawi. Int J Tuberc
Lung Dis 8 (2):211-7.
VanItallie, T. B., M. U. Yang, S. B. Heymsfield, R. C. Funk, and R. A. Boileau. 1990.
Height-normalized indices of the body's fat-free mass and fat mass: potentially
useful indicators of nutritional status. Am J Clin Nutr 52 (6):953-9.
Villamor, E., E. Saathoff, F. Mugusi, R. J. Bosch, W. Urassa, and W. W. Fawzi. 2006.
Wasting and body composition of adults with pulmonary tuberculosis in relation
to HIV-1 coinfection, socioeconomic status, and severity of tuberculosis. Eur J
Clin Nutr 60 (2):163-71.
454
CHAPTER FOUR
Bisai, S., and K. Bose. 2009. Undernutrition in the Kora Mudi tribal population, West
Bengal, India: a comparison of body mass index and mid-upper-arm
circumference. Food Nutr Bull 30 (1):63-7.
Blaak, E. 2001. Gender differences in fat metabolism. Curr Opin Clin Nutr Metab Care 4
(6):499-502.
Bland, J. M., and D. G. Altman. 1986. Statistical methods for assessing agreement
between two methods of clinical measurement. Lancet 1 (8476):307-10.
Briend, A., M. Garenne, B. Maire, O. Fontaine, and K. Dieng. 1989. Nutritional status,
age and survival: the muscle mass hypothesis. Eur J Clin Nutr 43 (10):715-26.
Collins, S. 1996. Using middle upper arm circumference to assess severe adult
malnutrition during famine. Jama 276 (5):391-5.
Dioum, A., A. Gartner, B. Maire, F. Delpeuch, and S. Wade. 2005. Body composition
predicted from skinfolds in African women: a cross-validation study using air-
displacement plethysmography and a black-specific equation. Br J Nutr 93
(6):973-9.
Durnin, J. V., and J. Womersley. 1974. Body fat assessed from total body density and its
estimation from skinfold thickness: measurements on 481 men and women aged
from 16 to 72 years. Br J Nutr 32 (1):77-97.
455
Flegal, K. M., J. A. Shepherd, A. C. Looker, B. I. Graubard, L. G. Borrud, C. L. Ogden,
T. B. Harris, J. E. Everhart, and N. Schenker. 2009. Comparisons of percentage
body fat, body mass index, waist circumference, and waist-stature ratio in adults.
Am J Clin Nutr 89 (2):500-8.
Frisancho A.R, ed. 1990. Anthropometric Standards for the Assessment of Growth and
Nutritional Status. Ann Arbor: The University of Michigan Press.
Harries, A. D., W. A. Nkhoma, P. J. Thompson, D. S. Nyangulu, and J. J. Wirima. 1988.
Nutritional status in Malawian patients with pulmonary tuberculosis and response
to chemotherapy. Eur J Clin Nutr 42 (5):445-50.
Heitmann, B. L., H. Erikson, B. M. Ellsinger, K. L. Mikkelsen, and B. Larsson. 2000.
Mortality associated with body fat, fat-free mass and body mass index among 60-
year-old swedish men-a 22-year follow-up. The study of men born in 1913. Int J
Obes Relat Metab Disord 24 (1):33-7.
James, W. P., G. C. Mascie-Taylor, N. G. Norgan, B. R. Bistrian, P. S. Shetty, and A.
Ferro-Luzzi. 1994. The value of arm circumference measurements in assessing
chronic energy deficiency in Third World adults. Eur J Clin Nutr 48 (12):883-94.
Kennedy, N., A. Ramsay, L. Uiso, J. Gutmann, F. I. Ngowi, and S. H. Gillespie. 1996.
Nutritional status and weight gain in patients with pulmonary tuberculosis in
Tanzania. Trans R Soc Trop Med Hyg 90 (2):162-6.
456
Kotler, D. P., S. Burastero, J. Wang, and R. N. Pierson, Jr. 1996. Prediction of body cell
mass, fat-free mass, and total body water with bioelectrical impedance analysis:
effects of race, sex, and disease. Am J Clin Nutr 64 (3 Suppl):489S-497S.
Kyle, U. G., I. Bosaeus, A. D. De Lorenzo, P. Deurenberg, M. Elia, J. M. Gomez, B. L.
Heitmann, L. Kent-Smith, J. C. Melchior, M. Pirlich, H. Scharfetter, A. M.
Schols, and C. Pichard. 2004. Bioelectrical impedance analysis--part I: review of
principles and methods. Clin Nutr 23 (5):1226-43.
Kyle, U. G., I. Bosaeus, A. D. De Lorenzo, P. Deurenberg, M. Elia, J. Manuel Gomez, B.
Lilienthal Heitmann, L. Kent-Smith, J. C. Melchior, M. Pirlich, H. Scharfetter, M.
W. J. Schols A, and C. Pichard. 2004. Bioelectrical impedance analysis-part II:
utilization in clinical practice. Clin Nutr 23 (6):1430-53.
Kyle, U. G., L. Genton, and C. Pichard. 2002. Body composition: what's new? Curr Opin
Clin Nutr Metab Care 5 (4):427-33.
Kyle, U. G., A. Piccoli, and C. Pichard. 2003. Body composition measurements:
interpretation finally made easy for clinical use. Curr Opin Clin Nutr Metab Care
6 (4):387-93.
Lawn, S. D., and G. Churchyard. 2009. Epidemiology of HIV-associated tuberculosis.
Curr Opin HIV AIDS 4 (4):325-33.
Lean, M. E., T. S. Han, and P. Deurenberg. 1996. Predicting body composition by
densitometry from simple anthropometric measurements. Am J Clin Nutr 63
(1):4-14.
457
Lohman T.G, Roche A.F, and Martorell R, eds. 1988. Anthropometric Standardization
Reference Manual, Human Kinetics Books. Illinois: A Division of Human
Kinetics, Inc.
Lucas, S. B., K. M. De Cock, A. Hounnou, C. Peacock, M. Diomande, M. Honde, A.
Beaumel, L. Kestens, and A. Kadio. 1994. Contribution of tuberculosis to slim
disease in Africa. Bmj 308 (6943):1531-3.
Macallan, D. C. 1999. Malnutrition in tuberculosis. Diagn Microbiol Infect Dis 34
(2):153-7.
Mostert, R., A. Goris, C. Weling-Scheepers, E. F. Wouters, and A. M. Schols. 2000.
Tissue depletion and health related quality of life in patients with chronic
obstructive pulmonary disease. Respir Med 94 (9):859-67.
Mupere, E., S. Zalwango, A. Chiunda, A. Okwera, R. Mugerwa, and C. Whalen. 2010.
Body composition among HIV-seropositive and HIV-seronegative adult patients
with pulmonary tuberculosis in Uganda. Ann Epidemiol 20 (3):210-6.
Niyongabo, T., D. Henzel, M. Idi, S. Nimubona, E. Gikoro, J. C. Melchior, S. Matheron,
G. Kamanfu, B. Samb, B. Messing, J. Begue, P. Aubry, and B. Larouze. 1999.
Tuberculosis, human immunodeficiency virus infection, and malnutrition in
Burundi. Nutrition 15 (4):289-93.
Norgan, N. G. 2005. Laboratory and field measurements of body composition. Public
Health Nutr 8 (7A):1108-22.
458
Oosthuizen, G. M., G. Joubert, W. F. Mollentze, and E. Rosslee. 1997. Body fat
estimation in black South Africans: a pilot study. Cent Afr J Med 43 (5):126-31.
Paton, N. I., and Y. M. Ng. 2006. Body composition studies in patients with wasting
associated with tuberculosis. Nutrition 22 (3):245-51.
Powell-Tuck, J., and E. M. Hennessy. 2003. A comparison of mid upper arm
circumference, body mass index and weight loss as indices of undernutrition in
acutely hospitalized patients. Clin Nutr 22 (3):307-12.
Schutz, Y., U. U. Kyle, and C. Pichard. 2002. Fat-free mass index and fat mass index
percentiles in Caucasians aged 18-98 y. Int J Obes Relat Metab Disord 26
(7):953-60.
Schwenk A, and Macallan D.C. 2000. Tuberculosis, malnutrition and wasting. Curr Opin
Clin Nutr Metab Care 3:285 - 91.
Shah, S., C. Whalen, D. P. Kotler, H. Mayanja, A. Namale, G. Melikian, R. Mugerwa,
and R. D. Semba. 2001. Severity of human immunodeficiency virus infection is
associated with decreased phase angle, fat mass and body cell mass in adults with
pulmonary tuberculosis infection in Uganda. J Nutr 131 (11):2843-7.
Siri W.E. 1961. Body composition from fluid spaces and density: analyses of methods.
Washington, DC: National Academy of Sciences.
Tang, A. M., J. Forrester, D. Spiegelman, T. A. Knox, E. Tchetgen, and S. L. Gorbach.
2002. Weight loss and survival in HIV-positive patients in the era of highly active
antiretroviral therapy. J Acquir Immune Defic Syndr 31 (2):230-6.
459
Taylor, R. W., D. Keil, E. J. Gold, S. M. Williams, and A. Goulding. 1998. Body mass
index, waist girth, and waist-to-hip ratio as indexes of total and regional adiposity
in women: evaluation using receiver operating characteristic curves. Am J Clin
Nutr 67 (1):44-9.
van Lettow, M., W. W. Fawzi, and R. D. Semba. 2003. Triple trouble: the role of
malnutrition in tuberculosis and human immunodeficiency virus co-infection.
Nutr Rev 61 (3):81-90.
Van Lettow, M., J. J. Kumwenda, A. D. Harries, C. C. Whalen, T. E. Taha, N.
Kumwenda, C. Kang'ombe, and R. D. Semba. 2004. Malnutrition and the severity
of lung disease in adults with pulmonary tuberculosis in Malawi. Int J Tuberc
Lung Dis 8 (2):211-7.
VanItallie, T. B., M. U. Yang, S. B. Heymsfield, R. C. Funk, and R. A. Boileau. 1990.
Height-normalized indices of the body's fat-free mass and fat mass: potentially
useful indicators of nutritional status. Am J Clin Nutr 52 (6):953-9.
Villamor, E., E. Saathoff, F. Mugusi, R. J. Bosch, W. Urassa, and W. W. Fawzi. 2006.
Wasting and body composition of adults with pulmonary tuberculosis in relation
to HIV-1 coinfection, socioeconomic status, and severity of tuberculosis. Eur J
Clin Nutr 60 (2):163-71.
Wagner, G. J., S. J. Ferrando, and J. G. Rabkin. 2000. Psychological and physical health
correlates of body cell mass depletion among HIV+ men. J Psychosom Res 49
(1):55-7.
460
Zachariah, R., M. P. Spielmann, A. D. Harries, and F. M. Salaniponi. 2002. Moderate to
severe malnutrition in patients with tuberculosis is a risk factor associated with
early death. Trans R Soc Trop Med Hyg 96 (3):291-4.
461
CHAPTER FIVE
Altman D.G, ed. 1997. Practical Statistics for Medical Research. London: Chapman &
Hall.
Atkinson, S. A., and W. E. Ward. 2001. Clinical nutrition: 2. The role of nutrition in the
prevention and treatment of adult osteoporosis. CMAJ 165 (11):1511-4.
Biro, G., K. F. Hulshof, L. Ovesen, and J. A. Amorim Cruz. 2002. Selection of
methodology to assess food intake. Eur J Clin Nutr 56 Suppl 2:S25-32.
Cannon G. 2001. Diet-related chronic diseases. Focus 5 (brief 8 of 11):1 - 2.
Carloni, A. S. 1981. Sex disparities in the distribution of food within rural households.
Food Nutr (Roma) 7 (1):3-12. de Hartog A.P. 1972. Unequal distribution of food within the household: A somewhat
neglected aspect of food behavior. FAO Nutritional Newletter 10; 8 - 17.
Dey J. 1981. Gambian women: Unequal partners in rice development projects? In African
Women in the Development Process, edited by Nelson N. London: Frank Cass.
Dietary Reference Intake for Energy, Carbohydrate, Fiber, Fat, Fatty Acids, Cholesterol,
Protein, and Amino Acids (Macronutrients). A report of the Panel on
Macronutrients, Subcommittees on Upper Reference Levels of Nutrients and
Interpretation and Uses of Dietary Reference Intakes, and the Standing
462
Committee on the Scientific Evaluation of Dietary Reference Intakes. 2005.
Washington, DC: National Academies Press. www.nap.edu.
Food and Agriculture Organization (FAO)/World Health Organization (WHO). Human
Vitamin and Mineral Requirements. Report of a Joint FAO/WHO Expert
Consultation. 2002. Rome.
Food and Nutrition Board, National Research Council, National Academy of Sciences.
Recommended Dietary Allowances. 1989. 10th ed. Washington, DC: National
Academic Press.
Ghafoorunissa K.K. 1994. Diet and Heart Disease. Hyderabad: National Institute of
Nutrition.
Gopalan C, Rama Sastri BV, and Basasubramanian S.C. 1996. Nutritive Value of Indian
Foods. Hyderabad: National Institute of Nutrition, Indian Council of Medical
Research.
Habicht, J. P., L. D. Meyers, and C. Brownie. 1982. Indicators for identifying and
counting the improperly nourished. Am J Clin Nutr 35 (5 Suppl):1241-54.
Hatloy, A., L. E. Torheim, and A. Oshaug. 1998. Food variety--a good indicator of
nutritional adequacy of the diet? A case study from an urban area in Mali, West
Africa. Eur J Clin Nutr 52 (12):891-8.
Holmboe-Ottesen G, and Wandel M. 1991. Men's contribution to the food and nutritional
situation in the Tanzanian household. Ecology of Food and Nutrition 26:83 - 96.
463
Hopewell P.C. 1994. Overview of clinical tuberculosis. In Tuberculosis: Pathogenesis,
Protection and Control, edited by Bloom B.R. Washington D.C: ASM Press.
Lado, C. 1992. Female labour participation in agricultural production and the
implications for nutrition and health in rural Africa. Soc Sci Med 34 (7):789-807.
Madden, J. P., S. J. Goodman, and H. A. Guthrie. 1976. Validity of the 24-hr. recall.
Analysis of data obtained from elderly subjects. J Am Diet Assoc 68 (2):143-7.
O'Laughlin B. 1974. Mediation of contradiction: Why Mbum women do not eat chicken.
In Women, Culture and Society, edited by Rosaldo M.Z. and Lamphere L.
Stanford, CA: Stanford University Press.
Oldewage-Theron, W. H., and R. Kruger. 2008. Food variety and dietary diversity as
indicators of the dietary adequacy and health status of an elderly population in
Sharpeville, South Africa. J Nutr Elder 27 (1-2):101-33.
Paton, N. I., Y. K. Chua, A. Earnest, and C. B. Chee. 2004. Randomized controlled trial
of nutritional supplementation in patients with newly diagnosed tuberculosis and
wasting. Am J Clin Nutr 80 (2):460-5.
Rosenberg E.M. 1980. Demographic effects of sex-differential nutrition. In Nutritional
Anthropology: Contemporary Approaches to Diet and Culture, edited by Jerome
J.W., Kandel R.F. and Pelto G.H. Pleasantville, NY: Redgrave Publishing.
Ruel, M. T. 2003. Operationalizing dietary diversity: a review of measurement issues and
research priorities. J Nutr 133 (11 Suppl 2):3911S-3926S.
464
Sita-Lumbsden A, Lapthorn G, Swaminathan R, and Milburn H.J. 2007. Reactivation of
tuberculosis and vitamin D deficiency: the contribution of diet and exposure to
sunlight. Thorax 62 (11):1003 - 1007.
Steyn, N. P., J. H. Nel, G. Nantel, G. Kennedy, and D. Labadarios. 2006. Food variety
and dietary diversity scores in children: are they good indicators of dietary
adequacy? Public Health Nutr 9 (5):644-50.
Thurnham, D. I. 2004. An overview of interactions between micronutrients and of
micronutrients with drugs, genes and immune mechanisms. Nutr Res Rev 17
(2):211-40.
Tontisirin, K., G. Nantel, and L. Bhattacharjee. 2002. Food-based strategies to meet the
challenges of micronutrient malnutrition in the developing world. Proc Nutr Soc
61 (2):243-50.
van Lettow, M., W. W. Fawzi, and R. D. Semba. 2003. Triple trouble: the role of
malnutrition in tuberculosis and human immunodeficiency virus co-infection.
Nutr Rev 61 (3):81-90.
Vucic, V., M. Glibetic, R. Novakovic, J. Ngo, D. Ristic-Medic, J. Tepsic, M. Ranic, L.
Serra-Majem, and M. Gurinovic. 2009. Dietary assessment methods used for low-
income populations in food consumption surveys: a literature review. Br J Nutr
101 Suppl 2:S95-101.
WHO. 2002. World health report 2002 - reducing risks, promoting healthy life. Geneva:
World Health Organization.
465
CHAPTER SIX
Arciero, P. J., M. I. Goran, and E. T. Poehlman. 1993. Resting metabolic rate is lower in
women than in men. J Appl Physiol 75 (6):2514-20.
Baumgartner, R. N., K. M. Koehler, D. Gallagher, L. Romero, S. B. Heymsfield, R. R.
Ross, P. J. Garry, and R. D. Lindeman. 1998. Epidemiology of sarcopenia among
the elderly in New Mexico. Am J Epidemiol 147 (8):755-63.
Corbett, E. L., C. J. Watt, N. Walker, D. Maher, B. G. Williams, M. C. Raviglione, and
C. Dye. 2003. The growing burden of tuberculosis: global trends and interactions
with the HIV epidemic. Arch Intern Med 163 (9):1009-21.
Dales, L. G., and H. K. Ury. 1978. An improper use of statistical significance testing in
studying covariables. Int J Epidemiol 7 (4):373-5.
Eckel, L. A. 2004. Estradiol: a rhythmic, inhibitory, indirect control of meal size. Physiol
Behav 82 (1):35-41.
Fields-Gardner, C. 1995. A review of mechanisms of wasting in HIV disease. Nutr Clin
Pract 10 (5):167-76.
Garrow, J. S., and J. Webster. 1985. Quetelet's index (W/H2) as a measure of fatness. Int
J Obes 9 (2):147-53.
Geary, N. 2001. Estradiol, CCK and satiation. Peptides 22 (8):1251-63.
Geary, N. 2001. Sex differences in disease anorexia. Nutrition 17 (6):499-507.
466
Kennedy, N., A. Ramsay, L. Uiso, J. Gutmann, F. I. Ngowi, and S. H. Gillespie. 1996.
Nutritional status and weight gain in patients with pulmonary tuberculosis in
Tanzania. Trans R Soc Trop Med Hyg 90 (2):162-6.
Kotler, D. P. 2000. Nutritional alterations associated with HIV infection. J Acquir
Immune Defic Syndr 25 Suppl 1:S81-7.
Kotler, D. P., S. Burastero, J. Wang, and R. N. Pierson, Jr. 1996. Prediction of body cell
mass, fat-free mass, and total body water with bioelectrical impedance analysis:
effects of race, sex, and disease. Am J Clin Nutr 64 (3 Suppl):489S-497S.
Kotler, D. P., and C. Grunfeld. 1995. Pathophysiology and treatment of the AIDS wasting
syndrome. AIDS Clin Rev:229-75.
Kyle, U. G., I. Bosaeus, A. D. De Lorenzo, P. Deurenberg, M. Elia, J. Manuel Gomez, B.
Lilienthal Heitmann, L. Kent-Smith, J. C. Melchior, M. Pirlich, H. Scharfetter, M.
W. J. Schols A, and C. Pichard. 2004. Bioelectrical impedance analysis-part II:
utilization in clinical practice. Clin Nutr 23 (6):1430-53.
Kyle, U. G., L. Genton, and C. Pichard. 2002. Body composition: what's new? Curr Opin
Clin Nutr Metab Care 5 (4):427-33.
Kyle, U. G., A. Piccoli, and C. Pichard. 2003. Body composition measurements:
interpretation finally made easy for clinical use. Curr Opin Clin Nutr Metab Care
6 (4):387-93.
467
Kyle, U. G., Y. Schutz, Y. M. Dupertuis, and C. Pichard. 2003. Body composition
interpretation. Contributions of the fat-free mass index and the body fat mass
index. Nutrition 19 (7-8):597-604.
Lawn, S. D., and G. Churchyard. 2009. Epidemiology of HIV-associated tuberculosis.
Curr Opin HIV AIDS 4 (4):325-33.
Lennie, T. A. 2004. Sex differences in severity of inflammation-induced anorexia and
weight loss. Biol Res Nurs 5 (4):255-64.
Lucas, S. B., K. M. De Cock, A. Hounnou, C. Peacock, M. Diomande, M. Honde, A.
Beaumel, L. Kestens, and A. Kadio. 1994. Contribution of tuberculosis to slim
disease in Africa. Bmj 308 (6943):1531-3.
Macallan, D. C. 1999. Malnutrition in tuberculosis. Diagn Microbiol Infect Dis 34
(2):153-7.
Macallan, D. C., M. A. McNurlan, A. V. Kurpad, G. de Souza, P. S. Shetty, A. G. Calder,
and G. E. Griffin. 1998. Whole body protein metabolism in human pulmonary
tuberculosis and undernutrition: evidence for anabolic block in tuberculosis. Clin
Sci (Lond) 94 (3):321-31.
Mupere, E., S. Zalwango, A. Chiunda, A. Okwera, R. Mugerwa, and C. Whalen. 2010.
Body composition among HIV-seropositive and HIV-seronegative adult patients
with pulmonary tuberculosis in Uganda. Ann Epidemiol 20 (3):210-6.
468
Neter J., Wasserman W., and Kutner M.H. 1990. Applied linear statistical models:
regression, analysis of variance, and experimental designs. 3rd ed. Homewood,
IL: Richard D Irwin Inc.
Niyongabo, T., D. Henzel, M. Idi, S. Nimubona, E. Gikoro, J. C. Melchior, S. Matheron,
G. Kamanfu, B. Samb, B. Messing, J. Begue, P. Aubry, and B. Larouze. 1999.
Tuberculosis, human immunodeficiency virus infection, and malnutrition in
Burundi. Nutrition 15 (4):289-93.
Paton, N. I., L. R. Castello-Branco, G. Jennings, M. B. Ortigao-de-Sampaio, M. Elia, S.
Costa, and G. E. Griffin. 1999. Impact of tuberculosis on the body composition of
HIV-infected men in Brazil. J Acquir Immune Defic Syndr Hum Retrovirol 20
(3):265-71.
Paton, N. I., and Y. M. Ng. 2006. Body composition studies in patients with wasting
associated with tuberculosis. Nutrition 22 (3):245-51.
Paton, N. I., Y. M. Ng, C. B. Chee, C. Persaud, and A. A. Jackson. 2003. Effects of
tuberculosis and HIV infection on whole-body protein metabolism during feeding,
measured by the [15N]glycine method. Am J Clin Nutr 78 (2):319-25.
Rammohan, M., K. Kalantar-Zadeh, A. Liang, and C. Ghossein. 2005. Megestrol acetate
in a moderate dose for the treatment of malnutrition-inflammation complex in
maintenance dialysis patients. J Ren Nutr 15 (3):345-55.
Rubin, S. A. 1995. Tuberculosis. Captain of all these men of death. Radiol Clin North Am
33 (4):619-39.
469
Shah, S., C. Whalen, D. P. Kotler, H. Mayanja, A. Namale, G. Melikian, R. Mugerwa,
and R. D. Semba. 2001. Severity of human immunodeficiency virus infection is
associated with decreased phase angle, fat mass and body cell mass in adults with
pulmonary tuberculosis infection in Uganda. J Nutr 131 (11):2843-7.
Tarnopolsky, M. A. 2000. Gender differences in metabolism; nutrition and supplements.
J Sci Med Sport 3 (3):287-98.
Tarnopolsky, M. A. 2000. Gender differences in substrate metabolism during endurance exercise. Can J Appl Physiol 25 (4):312-27.
Tarnopolsky, M. A. 2008. Sex differences in exercise metabolism and the role of 17-beta
estradiol. Med Sci Sports Exerc 40 (4):648-54.
Tarnopolsky, M. A., and B. C. Ruby. 2001. Sex differences in carbohydrate metabolism.
Curr Opin Clin Nutr Metab Care 4 (6):521-6.
Van Lettow, M., J. J. Kumwenda, A. D. Harries, C. C. Whalen, T. E. Taha, N.
Kumwenda, C. Kang'ombe, and R. D. Semba. 2004. Malnutrition and the severity
of lung disease in adults with pulmonary tuberculosis in Malawi. Int J Tuberc
Lung Dis 8 (2):211-7.
VanItallie, T. B., M. U. Yang, S. B. Heymsfield, R. C. Funk, and R. A. Boileau. 1990.
Height-normalized indices of the body's fat-free mass and fat mass: potentially
useful indicators of nutritional status. Am J Clin Nutr 52 (6):953-9.
Villamor, E., E. Saathoff, F. Mugusi, R. J. Bosch, W. Urassa, and W. W. Fawzi. 2006.
Wasting and body composition of adults with pulmonary tuberculosis in relation
470
to HIV-1 coinfection, socioeconomic status, and severity of tuberculosis. Eur J
Clin Nutr 60 (2):163-71.
Vittinghoff E., Shiboski S.C., Glidden D.V, and McCulloch C.E. 2005. Regreession
Methods in Biostatistics, Linear, Logistic, Survival, and Repeated Measure
Models. Edited by Gail M., Krickeberg K., Wong W., Samet J. and Tsiatis A.
New York, NY: Springer Science + Business Media.
World Health Organization. Global tuberculosis control. Epidemiology, strategy,
financing. WHO/HTM/TB/2009.411. 2009. Geneva: World Health Organization
471
CHAPTER SEVEN
Arciero, P. J., M. I. Goran, and E. T. Poehlman. 1993. Resting metabolic rate is lower in
women than in men. J Appl Physiol 75 (6):2514-20.
Baumgartner, R. N., K. M. Koehler, D. Gallagher, L. Romero, S. B. Heymsfield, R. R.
Ross, P. J. Garry, and R. D. Lindeman. 1998. Epidemiology of sarcopenia among
the elderly in New Mexico. Am J Epidemiol 147 (8):755-63.
Carloni, A. S. 1981. Sex disparities in the distribution of food within rural households.
Food Nutr (Roma) 7 (1):3-12. de Hartog A.P. 1972. Unequal distribution of food within the household: A somewhat
neglected aspect of food behavior. FAO Nutritional Newletter 10; 8 - 17.
Dey J. 1981. Gambian women: Unequal partners in rice development projects? In African
Women in the Development Process, edited by Nelson N. London: Frank Cass.
Duarte, E. C., A. L. Bierrenbach, J. Barbosa da Silva, Jr., P. L. Tauil, and E. de Fatima
Duarte. 2009. Factors associated with deaths among pulmonary tuberculosis
patients: a case-control study with secondary data. J Epidemiol Community
Health 63 (3):233-8.
Heitmann, B. L., H. Erikson, B. M. Ellsinger, K. L. Mikkelsen, and B. Larsson. 2000.
Mortality associated with body fat, fat-free mass and body mass index among 60-
year-old swedish men-a 22-year follow-up. The study of men born in 1913. Int J
Obes Relat Metab Disord 24 (1):33-7.
472
Holmboe-Ottesen G, and Wandel M. 1991. Men's contribution to the food and nutritional
situation in the Tanzanian household. Ecology of Food and Nutrition 26:83 - 96.
Khan, A., T. R. Sterling, R. Reves, A. Vernon, and C. R. Horsburgh. 2006. Lack of
weight gain and relapse risk in a large tuberculosis treatment trial. Am J Respir
Crit Care Med 174 (3):344-8.
Kotler, D. P., S. Burastero, J. Wang, and R. N. Pierson, Jr. 1996. Prediction of body cell
mass, fat-free mass, and total body water with bioelectrical impedance analysis:
effects of race, sex, and disease. Am J Clin Nutr 64 (3 Suppl):489S-497S.
Kyle, U. G., L. Genton, and C. Pichard. 2002. Body composition: what's new? Curr Opin
Clin Nutr Metab Care 5 (4):427-33.
Kyle, U. G., A. Piccoli, and C. Pichard. 2003. Body composition measurements:
interpretation finally made easy for clinical use. Curr Opin Clin Nutr Metab Care
6 (4):387-93.
Lado, C. 1992. Female labour participation in agricultural production and the
implications for nutrition and health in rural Africa. Soc Sci Med 34 (7):789-807.
Lawn, S. D., and G. Churchyard. 2009. Epidemiology of HIV-associated tuberculosis.
Curr Opin HIV AIDS 4 (4):325-33.
Lucas, S. B., K. M. De Cock, A. Hounnou, C. Peacock, M. Diomande, M. Honde, A.
Beaumel, L. Kestens, and A. Kadio. 1994. Contribution of tuberculosis to slim
disease in Africa. Bmj 308 (6943):1531-3.
473
Macallan, D. C. 1999. Malnutrition in tuberculosis. Diagn Microbiol Infect Dis 34
(2):153-7.
Muller, O., and M. Krawinkel. 2005. Malnutrition and health in developing countries.
CMAJ 173 (3):279-86.
Mupere, E., S. Zalwango, A. Chiunda, A. Okwera, R. Mugerwa, and C. Whalen. 2010.
Body composition among HIV-seropositive and HIV-seronegative adult patients
with pulmonary tuberculosis in Uganda. Ann Epidemiol 20 (3):210-6.
O'Laughlin B. 1974. Mediation of contradiction: Why Mbum women do not eat chicken.
In Women, Culture and Society, edited by Rosaldo M.Z. and Lamphere L.
Stanford, CA: Stanford University Press.
Paton, N. I., Y. M. Ng, C. B. Chee, C. Persaud, and A. A. Jackson. 2003. Effects of
tuberculosis and HIV infection on whole-body protein metabolism during feeding,
measured by the [15N]glycine method. Am J Clin Nutr 78 (2):319-25.
Physical status: the use and interpretation of anthropometry. Report of a WHO Expert
Committee. World Health Organ Tech Rep Ser, 854: 1 - 452. 1995. Geneva.
Rosenberg E.M. 1980. Demographic effects of sex-differential nutrition. In Nutritional
Anthropology: Contemporary Approaches to Diet and Culture, edited by Jerome
J.W., Kandel R.F. and Pelto G.H. Pleasantville, NY: Redgrave Publishing.
Saraceni, V., B. S. King, S. C. Cavalcante, J. E. Golub, L. M. Lauria, L. H. Moulton, R.
E. Chaisson, and B. Durovni. 2008. Tuberculosis as primary cause of death
474
among AIDS cases in Rio de Janeiro, Brazil. Int J Tuberc Lung Dis 12 (7):769-
72.
Schutz, Y., U. U. Kyle, and C. Pichard. 2002. Fat-free mass index and fat mass index
percentiles in Caucasians aged 18-98 y. Int J Obes Relat Metab Disord 26
(7):953-60.
Shah, S., C. Whalen, D. P. Kotler, H. Mayanja, A. Namale, G. Melikian, R. Mugerwa,
and R. D. Semba. 2001. Severity of human immunodeficiency virus infection is
associated with decreased phase angle, fat mass and body cell mass in adults with
pulmonary tuberculosis infection in Uganda. J Nutr 131 (11):2843-7.
Swaminathan, S., C. Padmapriyadarsini, B. Sukumar, S. Iliayas, S. R. Kumar, C. Triveni,
P. Gomathy, B. Thomas, M. Mathew, and P. R. Narayanan. 2008. Nutritional
status of persons with HIV infection, persons with HIV infection and tuberculosis,
and HIV-negative individuals from southern India. Clin Infect Dis 46 (6):946-9.
Van Lettow, M., J. J. Kumwenda, A. D. Harries, C. C. Whalen, T. E. Taha, N.
Kumwenda, C. Kang'ombe, and R. D. Semba. 2004. Malnutrition and the severity
of lung disease in adults with pulmonary tuberculosis in Malawi. Int J Tuberc
Lung Dis 8 (2):211-7.
VanItallie, T. B., M. U. Yang, S. B. Heymsfield, R. C. Funk, and R. A. Boileau. 1990.
Height-normalized indices of the body's fat-free mass and fat mass: potentially
useful indicators of nutritional status. Am J Clin Nutr 52 (6):953-9.
475
Villamor, E., E. Saathoff, F. Mugusi, R. J. Bosch, W. Urassa, and W. W. Fawzi. 2006.
Wasting and body composition of adults with pulmonary tuberculosis in relation
to HIV-1 coinfection, socioeconomic status, and severity of tuberculosis. Eur J
Clin Nutr 60 (2):163-71.
Wejse, C., P. Gustafson, J. Nielsen, V. F. Gomes, P. Aaby, P. L. Andersen, and M.
Sodemann. 2008. TBscore: Signs and symptoms from tuberculosis patients in a
low-resource setting have predictive value and may be used to assess clinical
course. Scand J Infect Dis 40 (2):111-20.
Zachariah, R., M. P. Spielmann, A. D. Harries, and F. M. Salaniponi. 2002. Moderate to
severe malnutrition in patients with tuberculosis is a risk factor associated with
early death. Trans R Soc Trop Med Hyg 96 (3):291-4.
476
CHAPTER EIGHT
Atkinson, S. A., and W. E. Ward. 2001. Clinical nutrition: 2. The role of nutrition in the
prevention and treatment of adult osteoporosis. CMAJ 165 (11):1511-4.
Carloni, A. S. 1981. Sex disparities in the distribution of food within rural households.
Food Nutr (Roma) 7 (1):3-12.
Dales, L. G., and H. K. Ury. 1978. An improper use of statistical significance testing in
studying covariables. Int J Epidemiol 7 (4):373-5. de Hartog A.P. 1972. Unequal distribution of food within the household: A somewhat
neglected aspect of food behavior. FAO Nutritional Newletter 10; 8 - 17.
Dey J. 1981. Gambian women: Unequal partners in rice development projects? In African
Women in the Development Process, edited by Nelson N. London: Frank Cass.
Dietary Reference Intake for Energy, Carbohydrate, Fiber, Fat, Fatty Acids, Cholesterol,
Protein, and Amino Acids (Macronutrients). A report of the Panel on
Macronutrients, Subcommittees on Upper Reference Levels of Nutrients and
Interpretation and Uses of Dietary Reference Intakes, and the Standing
Committee on the Scientific Evaluation of Dietary Reference Intakes. 2005.
Washington, DC: National Academies Press. www.nap.edu.
Food and Agriculture Organization (FAO)/World Health Organization (WHO). Human
Vitamin and Mineral Requirements. Report of a Joint FAO/WHO Expert
Consultation. 2002. Rome.
477
Frieden, T. R., T. R. Sterling, S. S. Munsiff, C. J. Watt, and C. Dye. 2003. Tuberculosis.
Lancet 362 (9387):887-99.
Haller, L., R. Sossouhounto, I. M. Coulibaly, M. Dosso, M. Kone, H. Adom, U. A.
Meyer, B. Betschart, M. Wenk, W. E. Haefeli, L. R. Lobognon, M. Porquet, G.
Kabore, F. Sorenson, R. Reber-Liske, and D. Sturchler. 1999. Isoniazid plus
sulphadoxine-pyrimethamine can reduce morbidity of HIV-positive patients
treated for tuberculosis in Africa: a controlled clinical trial. Chemotherapy 45
(6):452-65.
Harries, A. D., N. J. Hargreaves, J. Kemp, A. Jindani, D. A. Enarson, D. Maher, and F.
M. Salaniponi. 2001. Deaths from tuberculosis in sub-Saharan African countries
with a high prevalence of HIV-1. Lancet 357 (9267):1519-23.
Harries, A. D., W. A. Nkhoma, P. J. Thompson, D. S. Nyangulu, and J. J. Wirima. 1988.
Nutritional status in Malawian patients with pulmonary tuberculosis and response
to chemotherapy. Eur J Clin Nutr 42 (5):445-50.
Holmboe-Ottesen G, and Wandel M. 1991. Men's contribution to the food and nutritional
situation in the Tanzanian household. Ecology of Food and Nutrition 26:83 - 96.
Lado, C. 1992. Female labour participation in agricultural production and the
implications for nutrition and health in rural Africa. Soc Sci Med 34 (7):789-807.
Lawn, S. D., and G. Churchyard. 2009. Epidemiology of HIV-associated tuberculosis.
Curr Opin HIV AIDS 4 (4):325-33.
478
Macallan, D. C., M. A. McNurlan, A. V. Kurpad, G. de Souza, P. S. Shetty, A. G. Calder,
and G. E. Griffin. 1998. Whole body protein metabolism in human pulmonary
tuberculosis and undernutrition: evidence for anabolic block in tuberculosis. Clin
Sci (Lond) 94 (3):321-31.
Mehta J.B, Fields C.L, Byrd R.P.J, and Roy T.M. 1996. Nutritional status and mortality
in respiratory failure caused by tuberculosis. Tenn Med 89:349-71.
Mickey, R. M., and S. Greenland. 1989. The impact of confounder selection criteria on
effect estimation. Am J Epidemiol 129 (1):125-37.
Mitnick, C., J. Bayona, E. Palacios, S. Shin, J. Furin, F. Alcantara, E. Sanchez, M. Sarria,
M. Becerra, M. C. Fawzi, S. Kapiga, D. Neuberg, J. H. Maguire, J. Y. Kim, and P.
Farmer. 2003. Community-based therapy for multidrug-resistant tuberculosis in
Lima, Peru. N Engl J Med 348 (2):119-28.
O'Laughlin B. 1974. Mediation of contradiction: Why Mbum women do not eat chicken.
In Women, Culture and Society, edited by Rosaldo M.Z. and Lamphere L.
Stanford, CA: Stanford University Press.
Paton, N. I., L. R. Castello-Branco, G. Jennings, M. B. Ortigao-de-Sampaio, M. Elia, S.
Costa, and G. E. Griffin. 1999. Impact of tuberculosis on the body composition of
HIV-infected men in Brazil. J Acquir Immune Defic Syndr Hum Retrovirol 20
(3):265-71.
479
Paton, N. I., Y. M. Ng, C. B. Chee, C. Persaud, and A. A. Jackson. 2003. Effects of
tuberculosis and HIV infection on whole-body protein metabolism during feeding,
measured by the [15N]glycine method. Am J Clin Nutr 78 (2):319-25.
Rao, V. K., E. P. Iademarco, V. J. Fraser, and M. H. Kollef. 1998. The impact of
comorbidity on mortality following in-hospital diagnosis of tuberculosis. Chest
114 (5):1244-52.
Rosenberg E.M. 1980. Demographic effects of sex-differential nutrition. In Nutritional
Anthropology: Contemporary Approaches to Diet and Culture, edited by Jerome
J.W., Kandel R.F. and Pelto G.H. Pleasantville, NY: Redgrave Publishing.
SAS Institute. 2003. The GEN MOD Procedure. In SAS/STAT User's Guide, Version
9.1.3. Cary, NC: SAS Institute.
Wacholder, S. 1986. Binomial regression in GLIM: estimating risk ratios and risk
differences. Am J Epidemiol 123 (1):174-84.
Zachariah, R., M. P. Spielmann, A. D. Harries, and F. M. Salaniponi. 2002. Moderate to
severe malnutrition in patients with tuberculosis is a risk factor associated with
early death. Trans R Soc Trop Med Hyg 96 (3):291-4.
480
CHAPTER NINE
Allison, D. B., M. S. Faith, M. Heo, and D. P. Kotler. 1997. Hypothesis concerning the
U-shaped relation between body mass index and mortality. Am J Epidemiol 146
(4):339-49.
Corbett, E. L., C. J. Watt, N. Walker, D. Maher, B. G. Williams, M. C. Raviglione, and
C. Dye. 2003. The growing burden of tuberculosis: global trends and interactions
with the HIV epidemic. Arch Intern Med 163 (9):1009-21.
Garrow, J. S., and J. Webster. 1985. Quetelet's index (W/H2) as a measure of fatness. Int
J Obes 9 (2):147-53.
Guwatudde, D., M. Nakakeeto, E. C. Jones-Lopez, A. Maganda, A. Chiunda, R. D.
Mugerwa, J. J. Ellner, G. Bukenya, and C. C. Whalen. 2003. Tuberculosis in
household contacts of infectious cases in Kampala, Uganda. Am J Epidemiol 158
(9):887-98.
Harries, A. D., W. A. Nkhoma, P. J. Thompson, D. S. Nyangulu, and J. J. Wirima. 1988.
Nutritional status in Malawian patients with pulmonary tuberculosis and response
to chemotherapy. Eur J Clin Nutr 42 (5):445-50.
Hernandez-Pando, R., H. Orozco, and D. Aguilar. 2009. Factors that deregulate the
protective immune response in tuberculosis. Arch Immunol Ther Exp (Warsz) 57
(5):355-67.
481
Holmes, C. B., E. Losina, R. P. Walensky, Y. Yazdanpanah, and K. A. Freedberg. 2003.
Review of human immunodeficiency virus type 1-related opportunistic infections
in sub-Saharan Africa. Clin Infect Dis 36 (5):652-62.
Hosmer D.W.Jr, and Lemeshow S, eds. 1999. Applied Survival Analysis. Regression
Modeling of Time to Event Data. 1 ed. New York: John Wiley & Sons, Inc.
International Union Against Tuberculosis and Lung Disease. Technical guide for sputum
examination for tuberculosis by direct microscopy. 1986. Bull Int Union Tuberc
Lung Dis 61:1 - 16.
Jackson, A. S., M. L. Pollock, J. E. Graves, and M. T. Mahar. 1988. Reliability and
validity of bioelectrical impedance in determining body composition. J Appl
Physiol 64 (2):529-34.
Kaplan E.I, and Meier P. 1958. Nonparametric estimation from incomplete observations.
JASA 53:457 - 481.
Kennedy, N., A. Ramsay, L. Uiso, J. Gutmann, F. I. Ngowi, and S. H. Gillespie. 1996.
Nutritional status and weight gain in patients with pulmonary tuberculosis in
Tanzania. Trans R Soc Trop Med Hyg 90 (2):162-6.
Kotler, D. P., S. Burastero, J. Wang, and R. N. Pierson, Jr. 1996. Prediction of body cell
mass, fat-free mass, and total body water with bioelectrical impedance analysis:
effects of race, sex, and disease. Am J Clin Nutr 64 (3 Suppl):489S-497S.
Kyle, U. G., I. Bosaeus, A. D. De Lorenzo, P. Deurenberg, M. Elia, J. Manuel Gomez, B.
Lilienthal Heitmann, L. Kent-Smith, J. C. Melchior, M. Pirlich, H. Scharfetter, M.
482
W. J. Schols A, and C. Pichard. 2004. Bioelectrical impedance analysis-part II:
utilization in clinical practice. Clin Nutr 23 (6):1430-53.
Kyle, U. G., L. Genton, and C. Pichard. 2002. Body composition: what's new? Curr Opin
Clin Nutr Metab Care 5 (4):427-33.
Kyle, U. G., A. Piccoli, and C. Pichard. 2003. Body composition measurements:
interpretation finally made easy for clinical use. Curr Opin Clin Nutr Metab Care
6 (4):387-93.
Kyle, U. G., Y. Schutz, Y. M. Dupertuis, and C. Pichard. 2003. Body composition
interpretation. Contributions of the fat-free mass index and the body fat mass
index. Nutrition 19 (7-8):597-604.
Lucas, S. B., K. M. De Cock, A. Hounnou, C. Peacock, M. Diomande, M. Honde, A.
Beaumel, L. Kestens, and A. Kadio. 1994. Contribution of tuberculosis to slim
disease in Africa. Bmj 308 (6943):1531-3.
Lucas, S. B., A. Hounnou, C. Peacock, A. Beaumel, G. Djomand, J. M. N'Gbichi, K.
Yeboue, M. Honde, M. Diomande, C. Giordano, and et al. 1993. The mortality
and pathology of HIV infection in a west African city. Aids 7 (12):1569-79.
Luginaah, I. N., E. K. Yiridoe, and M. M. Taabazuing. 2005. From mandatory to
voluntary testing: balancing human rights, religious and cultural values, and
HIV/AIDS prevention in Ghana. Soc Sci Med 61 (8):1689-700.
Macallan, D. C. 1999. Malnutrition in tuberculosis. Diagn Microbiol Infect Dis 34
(2):153-7.
483
McGinn, T., P. C. Wyer, T. B. Newman, S. Keitz, R. Leipzig, and G. G. For. 2004. Tips
for learners of evidence-based medicine: 3. Measures of observer variability
(kappa statistic). CMAJ 171 (11):1369-73.
Melchior, J. C., G. Raguin, A. Boulier, E. Bouvet, D. Rigaud, S. Matheron, E. Casalino,
J. L. Vilde, F. Vachon, J. P. Coulaud, and et al. 1993. Resting energy expenditure
in human immunodeficiency virus-infected patients: comparison between patients
with and without secondary infections. Am J Clin Nutr 57 (5):614-9.
Mitnick, C., J. Bayona, E. Palacios, S. Shin, J. Furin, F. Alcantara, E. Sanchez, M. Sarria,
M. Becerra, M. C. Fawzi, S. Kapiga, D. Neuberg, J. H. Maguire, J. Y. Kim, and P.
Farmer. 2003. Community-based therapy for multidrug-resistant tuberculosis in
Lima, Peru. N Engl J Med 348 (2):119-28.
Mugusi, F. M., S. Mehta, E. Villamor, W. Urassa, E. Saathoff, R. J. Bosch, and W. W.
Fawzi. 2009. Factors associated with mortality in HIV-infected and uninfected
patients with pulmonary tuberculosis. BMC Public Health 9:409.
Mupere, E., S. Zalwango, A. Chiunda, A. Okwera, R. Mugerwa, and C. Whalen. 2010.
Body composition among HIV-seropositive and HIV-seronegative adult patients
with pulmonary tuberculosis in Uganda. Ann Epidemiol 20 (3):210-6.
Niyongabo, T., D. Henzel, M. Idi, S. Nimubona, E. Gikoro, J. C. Melchior, S. Matheron,
G. Kamanfu, B. Samb, B. Messing, J. Begue, P. Aubry, and B. Larouze. 1999.
Tuberculosis, human immunodeficiency virus infection, and malnutrition in
Burundi. Nutrition 15 (4):289-93.
484
Niyongabo, T., N. Mlika-Cabanne, T. Barihuta, D. Henzel, P. Aubry, and B. Larauze.
1994. Malnutrition, tuberculosis and HIV infection in Burundi. Aids 8 (6):851-2.
Nunn, P., R. Brindle, L. Carpenter, J. Odhiambo, K. Wasunna, R. Newnham, W. Githui,
S. Gathua, M. Omwega, and K. McAdam. 1992. Cohort study of human
immunodeficiency virus infection in patients with tuberculosis in Nairobi, Kenya.
Analysis of early (6-month) mortality. Am Rev Respir Dis 146 (4):849-54.
Paton, N. I., L. R. Castello-Branco, G. Jennings, M. B. Ortigao-de-Sampaio, M. Elia, S.
Costa, and G. E. Griffin. 1999. Impact of tuberculosis on the body composition of
HIV-infected men in Brazil. J Acquir Immune Defic Syndr Hum Retrovirol 20
(3):265-71.
Paton, N. I., and Y. M. Ng. 2006. Body composition studies in patients with wasting
associated with tuberculosis. Nutrition 22 (3):245-51.
Paton, N. I., Y. M. Ng, C. B. Chee, C. Persaud, and A. A. Jackson. 2003. Effects of
tuberculosis and HIV infection on whole-body protein metabolism during feeding,
measured by the [15N]glycine method. Am J Clin Nutr 78 (2):319-25.
Physical status: the use and interpretation of anthropometry. Report of a WHO Expert
Committee. World Health Organ Tech Rep Ser, 854: 1 - 452. 1995. Geneva.
Rao, V. K., E. P. Iademarco, V. J. Fraser, and M. H. Kollef. 1998. The impact of
comorbidity on mortality following in-hospital diagnosis of tuberculosis. Chest
114 (5):1244-52.
485
Sani, M. U., A. Z. Mohammed, B. Adamu, S. M. Yusuf, A. A. Samaila, and M. M.
Borodo. 2006. AIDS mortality in a tertiary health institution: A four-year review.
J Natl Med Assoc 98 (6):862-6.
Schutz, Y., U. U. Kyle, and C. Pichard. 2002. Fat-free mass index and fat mass index
percentiles in Caucasians aged 18-98 y. Int J Obes Relat Metab Disord 26
(7):953-60.
Schwenk A, and Macallan D.C. 2000. Tuberculosis, malnutrition and wasting. Curr Opin
Clin Nutr Metab Care 3:285 - 91.
Shah, S., C. Whalen, D. P. Kotler, H. Mayanja, A. Namale, G. Melikian, R. Mugerwa,
and R. D. Semba. 2001. Severity of human immunodeficiency virus infection is
associated with decreased phase angle, fat mass and body cell mass in adults with
pulmonary tuberculosis infection in Uganda. J Nutr 131 (11):2843-7.
Stein, C. M., L. Nshuti, A. B. Chiunda, W. H. Boom, R. C. Elston, R. D. Mugerwa, S. K.
Iyengar, and C. C. Whalen. 2005. Evidence for a major gene influence on tumor
necrosis factor-alpha expression in tuberculosis: path and segregation analysis.
Hum Hered 60 (2):109-18.
Suttmann, U., J. Ockenga, O. Selberg, L. Hoogestraat, H. Deicher, and M. J. Muller.
1995. Incidence and prognostic value of malnutrition and wasting in human
immunodeficiency virus-infected outpatients. J Acquir Immune Defic Syndr Hum
Retrovirol 8 (3):239-46.
486
van Lettow, M., W. W. Fawzi, and R. D. Semba. 2003. Triple trouble: the role of
malnutrition in tuberculosis and human immunodeficiency virus co-infection.
Nutr Rev 61 (3):81-90.
Van Lettow, M., J. J. Kumwenda, A. D. Harries, C. C. Whalen, T. E. Taha, N.
Kumwenda, C. Kang'ombe, and R. D. Semba. 2004. Malnutrition and the severity
of lung disease in adults with pulmonary tuberculosis in Malawi. Int J Tuberc
Lung Dis 8 (2):211-7.
VanItallie, T. B., M. U. Yang, S. B. Heymsfield, R. C. Funk, and R. A. Boileau. 1990.
Height-normalized indices of the body's fat-free mass and fat mass: potentially
useful indicators of nutritional status. Am J Clin Nutr 52 (6):953-9.
Villamor, E., E. Saathoff, F. Mugusi, R. J. Bosch, W. Urassa, and W. W. Fawzi. 2006.
Wasting and body composition of adults with pulmonary tuberculosis in relation
to HIV-1 coinfection, socioeconomic status, and severity of tuberculosis. Eur J
Clin Nutr 60 (2):163-71.
Vitoria, M., R. Granich, C. F. Gilks, C. Gunneberg, M. Hosseini, W. Were, M.
Raviglione, and K. M. De Cock. 2009. The global fight against HIV/AIDS,
tuberculosis, and malaria: current status and future perspectives. Am J Clin Pathol
131 (6):844-8.
Zachariah, R., M. P. Spielmann, A. D. Harries, and F. M. Salaniponi. 2002. Moderate to
severe malnutrition in patients with tuberculosis is a risk factor associated with
early death. Trans R Soc Trop Med Hyg 96 (3):291-4.
487
CHAPTER TEN
Blaak, E. 2001. Gender differences in fat metabolism. Curr Opin Clin Nutr Metab Care 4
(6):499-502.
Bryk A.S, and Ruandenbush S.W. 2002. Hierarchical linear models: applications and
data analysis methods. 2 ed. Thousands Oaks, CA Sage.
Chemotherapy and management of tuberculosis in the United Kingdom:
recommendations 1998. Joint Tuberculosis Committee of the British Thoracic
Society. 1998. Thorax 53 (7):536-48.
Ferro-Luzzi, A., C. Petracchi, R. Kuriyan, and A. V. Kurpad. 1997. Basal metabolism of
weight-stable chronically undernourished men and women: lack of metabolic
adaptation and ethnic differences. Am J Clin Nutr 66 (5):1086-93.
Guwatudde, D., M. Nakakeeto, E. C. Jones-Lopez, A. Maganda, A. Chiunda, R. D.
Mugerwa, J. J. Ellner, G. Bukenya, and C. C. Whalen. 2003. Tuberculosis in
household contacts of infectious cases in Kampala, Uganda. Am J Epidemiol 158
(9):887-98.
Harries, A. D., W. A. Nkhoma, P. J. Thompson, D. S. Nyangulu, and J. J. Wirima. 1988.
Nutritional status in Malawian patients with pulmonary tuberculosis and response
to chemotherapy. Eur J Clin Nutr 42 (5):445-50.
488
Horton, T. J., M. J. Pagliassotti, K. Hobbs, and J. O. Hill. 1998. Fuel metabolism in men
and women during and after long-duration exercise. J Appl Physiol 85 (5):1823-
32.
Jackson, A. S., M. L. Pollock, J. E. Graves, and M. T. Mahar. 1988. Reliability and
validity of bioelectrical impedance in determining body composition. J Appl
Physiol 64 (2):529-34.
Kennedy, N., A. Ramsay, L. Uiso, J. Gutmann, F. I. Ngowi, and S. H. Gillespie. 1996.
Nutritional status and weight gain in patients with pulmonary tuberculosis in
Tanzania. Trans R Soc Trop Med Hyg 90 (2):162-6.
Kotler, D. P., S. Burastero, J. Wang, and R. N. Pierson, Jr. 1996. Prediction of body cell
mass, fat-free mass, and total body water with bioelectrical impedance analysis:
effects of race, sex, and disease. Am J Clin Nutr 64 (3 Suppl):489S-497S.
Kurpad, A. V., S. Muthayya, and M. Vaz. 2005. Consequences of inadequate food energy
and negative energy balance in humans. Public Health Nutr 8 (7A):1053-76.
Kyle, U. G., I. Bosaeus, A. D. De Lorenzo, P. Deurenberg, M. Elia, J. Manuel Gomez, B.
Lilienthal Heitmann, L. Kent-Smith, J. C. Melchior, M. Pirlich, H. Scharfetter, M.
W. J. Schols A, and C. Pichard. 2004. Bioelectrical impedance analysis-part II:
utilization in clinical practice. Clin Nutr 23 (6):1430-53.
Kyle, U. G., L. Genton, and C. Pichard. 2002. Body composition: what's new? Curr Opin
Clin Nutr Metab Care 5 (4):427-33.
489
Kyle, U. G., A. Piccoli, and C. Pichard. 2003. Body composition measurements:
interpretation finally made easy for clinical use. Curr Opin Clin Nutr Metab Care
6 (4):387-93.
Kyle, U. G., Y. Schutz, Y. M. Dupertuis, and C. Pichard. 2003. Body composition
interpretation. Contributions of the fat-free mass index and the body fat mass
index. Nutrition 19 (7-8):597-604.
Laird, N. M., and J. H. Ware. 1982. Random-effects models for longitudinal data.
Biometrics 38 (4):963-74.
Littell, R. C., J. Pendergast, and R. Natarajan. 2000. Modelling covariance structure in
the analysis of repeated measures data. Stat Med 19 (13):1793-819.
Lucas, S. B., K. M. De Cock, A. Hounnou, C. Peacock, M. Diomande, M. Honde, A.
Beaumel, L. Kestens, and A. Kadio. 1994. Contribution of tuberculosis to slim
disease in Africa. Bmj 308 (6943):1531-3.
Macallan, D. C. 1999. Malnutrition in tuberculosis. Diagn Microbiol Infect Dis 34
(2):153-7.
Macallan, D. C., M. A. McNurlan, A. V. Kurpad, G. de Souza, P. S. Shetty, A. G. Calder,
and G. E. Griffin. 1998. Whole body protein metabolism in human pulmonary
tuberculosis and undernutrition: evidence for anabolic block in tuberculosis. Clin
Sci (Lond) 94 (3):321-31.
Mayanja-Kizza, H., E. Jones-Lopez, A. Okwera, R. S. Wallis, J. J. Ellner, R. D.
Mugerwa, and C. C. Whalen. 2005. Immunoadjuvant prednisolone therapy for
490
HIV-associated tuberculosis: a phase 2 clinical trial in Uganda. J Infect Dis 191
(6):856-65.
Mitnick, C., J. Bayona, E. Palacios, S. Shin, J. Furin, F. Alcantara, E. Sanchez, M. Sarria,
M. Becerra, M. C. Fawzi, S. Kapiga, D. Neuberg, J. H. Maguire, J. Y. Kim, and P.
Farmer. 2003. Community-based therapy for multidrug-resistant tuberculosis in
Lima, Peru. N Engl J Med 348 (2):119-28.
Mooradian, A. D., J. E. Morley, and S. G. Korenman. 1987. Biological actions of
androgens. Endocr Rev 8 (1):1-28.
Mostert, R., A. Goris, C. Weling-Scheepers, E. F. Wouters, and A. M. Schols. 2000.
Tissue depletion and health related quality of life in patients with chronic
obstructive pulmonary disease. Respir Med 94 (9):859-67.
Mupere, E., S. Zalwango, A. Chiunda, A. Okwera, R. Mugerwa, and C. Whalen. 2010.
Body composition among HIV-seropositive and HIV-seronegative adult patients
with pulmonary tuberculosis in Uganda. Ann Epidemiol 20 (3):210-6.
Nagy, T. R., M. I. Goran, R. L. Weinsier, M. J. Toth, Y. Schutz, and E. T. Poehlman.
1996. Determinants of basal fat oxidation in healthy Caucasians. J Appl Physiol
80 (5):1743-8.
Niyongabo, T., D. Henzel, M. Idi, S. Nimubona, E. Gikoro, J. C. Melchior, S. Matheron,
G. Kamanfu, B. Samb, B. Messing, J. Begue, P. Aubry, and B. Larouze. 1999.
Tuberculosis, human immunodeficiency virus infection, and malnutrition in
Burundi. Nutrition 15 (4):289-93.
491
Niyongabo, T., N. Mlika-Cabanne, T. Barihuta, D. Henzel, P. Aubry, and B. Larauze.
1994. Malnutrition, tuberculosis and HIV infection in Burundi. Aids 8 (6):851-2.
Onwubalili, J. K. 1988. Malnutrition among tuberculosis patients in Harrow, England.
Eur J Clin Nutr 42 (4):363-6.
Paton, N. I., B. Angus, W. Chaowagul, A. J. Simpson, Y. Suputtamongkol, M. Elia, G.
Calder, E. Milne, N. J. White, and G. E. Griffin. 2001. Protein and energy
metabolism in chronic bacterial infection: studies in melioidosis. Clin Sci (Lond)
100 (1):101-10.
Paton, N. I., L. R. Castello-Branco, G. Jennings, M. B. Ortigao-de-Sampaio, M. Elia, S.
Costa, and G. E. Griffin. 1999. Impact of tuberculosis on the body composition of
HIV-infected men in Brazil. J Acquir Immune Defic Syndr Hum Retrovirol 20
(3):265-71.
Paton, N. I., Y. K. Chua, A. Earnest, and C. B. Chee. 2004. Randomized controlled trial
of nutritional supplementation in patients with newly diagnosed tuberculosis and
wasting. Am J Clin Nutr 80 (2):460-5.
Paton, N. I., and Y. M. Ng. 2006. Body composition studies in patients with wasting
associated with tuberculosis. Nutrition 22 (3):245-51.
Paton, N. I., Y. M. Ng, C. B. Chee, C. Persaud, and A. A. Jackson. 2003. Effects of
tuberculosis and HIV infection on whole-body protein metabolism during feeding,
measured by the [15N]glycine method. Am J Clin Nutr 78 (2):319-25.
492
Physical status: the use and interpretation of anthropometry. Report of a WHO Expert
Committee. World Health Organ Tech Rep Ser, 854: 1 - 452. 1995. Geneva.
Ramakrishnan, C. V., K. Rajendran, P. G. Jacob, W. Fox, and S. Radhakrishna. 1961.
The role of diet in the treatment of pulmonary tuberculosis. An evaluation in a
controlled chemotherapy study in home and sanatorium patients in South India.
Bull World Health Organ 25:339-59.
Rao, V. K., E. P. Iademarco, V. J. Fraser, and M. H. Kollef. 1998. The impact of
comorbidity on mortality following in-hospital diagnosis of tuberculosis. Chest
114 (5):1244-52.
Ridout, M. S. 1991. Testing for random dropouts in repeated measurement data.
Biometrics 47 (4):1617-9; discussion 1619-21.
SAS Institute. The Mixed Procedure. In SAS/STAT User's Guide, Version 9.1.3. Cary,
NC: SAS Institute.
Schutz, Y., U. U. Kyle, and C. Pichard. 2002. Fat-free mass index and fat mass index
percentiles in Caucasians aged 18-98 y. Int J Obes Relat Metab Disord 26
(7):953-60.
Schwenk A, and Macallan D.C. 2000. Tuberculosis, malnutrition and wasting. Curr Opin
Clin Nutr Metab Care 3:285 - 91.
Schwenk, A., L. Hodgson, A. Wright, L. C. Ward, C. F. Rayner, S. Grubnic, G. E.
Griffin, and D. C. Macallan. 2004. Nutrient partitioning during treatment of
493
tuberculosis: gain in body fat mass but not in protein mass. Am J Clin Nutr 79
(6):1006-12.
Shah, S., C. Whalen, D. P. Kotler, H. Mayanja, A. Namale, G. Melikian, R. Mugerwa,
and R. D. Semba. 2001. Severity of human immunodeficiency virus infection is
associated with decreased phase angle, fat mass and body cell mass in adults with
pulmonary tuberculosis infection in Uganda. J Nutr 131 (11):2843-7.
Singer J.D, and Willet J.B. 2003. Applied longitudinal data analysis: modeling change
and event occurrence. 1 ed: Oxford University Press.
Stein, C. M., L. Nshuti, A. B. Chiunda, W. H. Boom, R. C. Elston, R. D. Mugerwa, S. K.
Iyengar, and C. C. Whalen. 2005. Evidence for a major gene influence on tumor
necrosis factor-alpha expression in tuberculosis: path and segregation analysis.
Hum Hered 60 (2):109-18.
Swanson, B., R. C. Hershow, B. E. Sha, C. A. Benson, M. Cohen, and C. Gunfeld. 2000.
Body composition in HIV-infected women. Nutrition 16 (11-12):1064-8.
Tomkins, A. M., P. J. Garlick, W. N. Schofield, and J. C. Waterlow. 1983. The combined
effects of infection and malnutrition on protein metabolism in children. Clin Sci
(Lond) 65 (3):313-24.
Toth, M. J., P. J. Arciero, A. W. Gardner, J. Calles-Escandon, and E. T. Poehlman. 1996.
Rates of free fatty acid appearance and fat oxidation in healthy younger and older
men. J Appl Physiol 80 (2):506-11.
494
Toth, M. J., A. W. Gardner, P. J. Arciero, J. Calles-Escandon, and E. T. Poehlman. 1998.
Gender differences in fat oxidation and sympathetic nervous system activity at
rest and during submaximal exercise in older individuals. Clin Sci (Lond) 95
(1):59-66.
van Lettow, M., W. W. Fawzi, and R. D. Semba. 2003. Triple trouble: the role of
malnutrition in tuberculosis and human immunodeficiency virus co-infection.
Nutr Rev 61 (3):81-90.
Van Lettow, M., J. J. Kumwenda, A. D. Harries, C. C. Whalen, T. E. Taha, N.
Kumwenda, C. Kang'ombe, and R. D. Semba. 2004. Malnutrition and the severity
of lung disease in adults with pulmonary tuberculosis in Malawi. Int J Tuberc
Lung Dis 8 (2):211-7.
VanItallie, T. B., M. U. Yang, S. B. Heymsfield, R. C. Funk, and R. A. Boileau. 1990.
Height-normalized indices of the body's fat-free mass and fat mass: potentially
useful indicators of nutritional status. Am J Clin Nutr 52 (6):953-9.
Villamor, E., E. Saathoff, F. Mugusi, R. J. Bosch, W. Urassa, and W. W. Fawzi. 2006.
Wasting and body composition of adults with pulmonary tuberculosis in relation
to HIV-1 coinfection, socioeconomic status, and severity of tuberculosis. Eur J
Clin Nutr 60 (2):163-71.
Vitoria, M., R. Granich, C. F. Gilks, C. Gunneberg, M. Hosseini, W. Were, M.
Raviglione, and K. M. De Cock. 2009. The global fight against HIV/AIDS,
tuberculosis, and malaria: current status and future perspectives. Am J Clin Pathol
131 (6):844-8.
495
Wagner, G. J., S. J. Ferrando, and J. G. Rabkin. 2000. Psychological and physical health
correlates of body cell mass depletion among HIV+ men. J Psychosom Res 49
(1):55-7.
Zachariah, R., M. P. Spielmann, A. D. Harries, and F. M. Salaniponi. 2002. Moderate to
severe malnutrition in patients with tuberculosis is a risk factor associated with
early death. Trans R Soc Trop Med Hyg 96 (3):291-4.
Zurlo, F., S. Lillioja, A. Esposito-Del Puente, B. L. Nyomba, I. Raz, M. F. Saad, B. A.
Swinburn, W. C. Knowler, C. Bogardus, and E. Ravussin. 1990. Low ratio of fat
to carbohydrate oxidation as predictor of weight gain: study of 24-h RQ. Am J
Physiol 259 (5 Pt 1):E650-7.
496
CHAPTER ELEVEN
Kyle, U. G., L. Genton, and C. Pichard. 2002. Body composition: what's new? Curr Opin
Clin Nutr Metab Care 5 (4):427-33.
Kyle, U. G., A. Piccoli, and C. Pichard. 2003. Body composition measurements:
interpretation finally made easy for clinical use. Curr Opin Clin Nutr Metab Care
6 (4):387-93.
Lucas, S. B., A. Hounnou, C. Peacock, A. Beaumel, G. Djomand, J. M. N'Gbichi, K.
Yeboue, M. Honde, M. Diomande, C. Giordano, and et al. 1993. The mortality
and pathology of HIV infection in a west African city. Aids 7 (12):1569-79.
Mupere, E., S. Zalwango, A. Chiunda, A. Okwera, R. Mugerwa, and C. Whalen. 2010.
Body composition among HIV-seropositive and HIV-seronegative adult patients
with pulmonary tuberculosis in Uganda. Ann Epidemiol 20 (3):210-6.
Sani, M. U., A. Z. Mohammed, B. Adamu, S. M. Yusuf, A. A. Samaila, and M. M.
Borodo. 2006. AIDS mortality in a tertiary health institution: A four-year review.
J Natl Med Assoc 98 (6):862-6.
Steyn, N. P., P. Wolmarans, J. H. Nel, and L. T. Bourne. 2008. National fortification of
staple foods can make a significant contribution to micronutrient intake of South
African adults. Public Health Nutr 11 (3):307-13.
VanItallie, T. B., M. U. Yang, S. B. Heymsfield, R. C. Funk, and R. A. Boileau. 1990.
Height-normalized indices of the body's fat-free mass and fat mass: potentially
useful indicators of nutritional status. Am J Clin Nutr 52 (6):953-9.
497
Vucic, V., M. Glibetic, R. Novakovic, J. Ngo, D. Ristic-Medic, J. Tepsic, M. Ranic, L.
Serra-Majem, and M. Gurinovic. 2009. Dietary assessment methods used for low-
income populations in food consumption surveys: a literature review. Br J Nutr
101 Suppl 2:S95-101.
498